Trouble with fillet around awkward shape

[Hend-Eng]

Hi,

Please see attached screenshot. 

Re-creating a 1920s front axle from drawings and for some reason F360 cannot produce a fillet around this bracket / cylinder intersection. 

I was previously able to do it before I added guide curves to give the rectagular bracket it's proper curvature, 200mm radius; with the previous 2D curved faces I could fillet sucessfully.

Are there any workarounds available?

Without the 3D curves of the rectangular bracket the finish machined item just wont look as it should.

Workflow was to make a 1/4 of the axle, sweep the half cylinder and loft the rectangular sections from 2 planes with guide rails in 2 planes to get the right shape then mirrored the section to create one half.

Rads required from R10 to R30.

[laughingcreek]

The lesson in all this, don't use a loft when a sweep will do.  the loft gave you decent geometry after a lot of work (multiple sketches  rails, profiles, helperbodys etc)

 and the sweep gives you great geometry with one profile, a rail a path and a rail-

[beresfordromeo]

---

@laughingcreek wrote:

The lesson in all this, don't use a loft when a sweep will do.  the loft gave you decent geometry after a lot of work (multiple sketches  rails, profiles, helperbodys etc)

 and the sweep gives you great geometry with one profile, a rail a path and a rail-

---

Hi @laughingcreek 

Whilst I don't disagree with you about 'the lesson in all this', to describe it this way is perhaps a little reductive and a little unfair to the Loft tool. I don't mean to be contrary or precocious, I understand that your comment is both shorthand and humorous but it seems worth iterating that the problem with the model was the dependency on the mirror and combine to replicate the second half of the body.

I am not a proffessional in this field so I may be wrong but in my oppinion, if @Hend-Eng had projected the smaller of the two profiles (Sketch 3) to a an offset plane and drawn the rail arcs between then, the loft could have been performed as a single body and would have resulted in a class A surface. It would have also been more efficient because not only would the helper body not have been necessary but the Sweep could have been a Pipe (requiring one less sketch) and there would have been no need for the Mirror or Combine. In further defence of the Loft tool, the advantage, were it necessary would be the ability to check Merge for the Tangent Edges setting resulting in better surfaces.

I could of course be really wrong about that, thanks again for your time.

[laughingcreek]

---

@beresfordromeo wrote:

... but it seems worth iterating that the problem with the model was the dependency on the mirror and combine to replicate the second half of the body.

---

The problem with the loft was the location of the rails.  placing them mid span between curve tangent points could never have produced a "class A" surface.  Even lofting between 3 profiles will still produce subprime surfacing  if you use the same rail location.  You'll only get a great loft result with this setup if you either leave the rails off, or place more at the ends of the fillets, which would tricky.  (I'm guessing at this point you don't believe me, so I would invite you to do some experimentation.  You'll see what I mean)

So i'll reiterate, the basic axiom in modeling is "less is more".  As applied to surfacing that means use the simplest tool that will achieve the desired result.  Will an extrude achieve the right shape?  if yes, use that.  if no, then will a sweep or revolve achieve the desired form?  if yes, use that.  if not, use a loft.  Loft is always the last tool of choice, but frequently the only one that will do the job. 

[Hend-Eng]

Hi @beresfordromeo @laughingcreek 

I must ask you to stop - You've both gone way way over and above the call of duty!!

Modelling practice though, I was always encouraged to do the least necessary and take advantage of the fact that symmetrical parts can usually be reduced down to a 1/4 and mirrored to produce the finished article. One solution as suggested would have been to go for a half model instead and the awkward geometry could have been avoided perhaps.

Interesting that lofts aren't that great by the sounds of things, but presumably with plenty of guide sketches of multiple planes, perhaps less so.

The one major highlight of all of this conversation is how limited my CAD knowledge / abilities are. I've traditionally only dealt with the more regular shapes and issues such as this have never really cropped up before, I'll certainly be taking this thread on-board the next time I approach a funny set of shapes.

I've still yet to go through all of your work @beresfordromeo I'm afraid, there's an awful lot to take in and it's a whole different approach to tackling a modelling issue than I'm used to.

I've a feeling I'll be keeping this tab open for a very long time. Lots to learn.

Lastly, I almost feel like a cheat for going with the easy option of just cleaning up the model with a different approach - sweep instead of loft, but as above, I will try and up my game by trying out the dirty approach.

Thanks again everyone for leaving me truly humbled at your huge efforts with this.

Tom.

P.S. Anyone able to doctor the haas pre next gen post to stop it outputting a G94? My 25 year old machine doesn't understand it!

[beresfordromeo]

---

@laughingcreek wrote:

---

@beresfordromeo wrote:

... but it seems worth iterating that the problem with the model was the dependency on the mirror and combine to replicate the second half of the body.

---

The problem with the loft was the location of the rails.  placing them mid span between curve tangent points could never have produced a "class A" surface.  Even lofting between 3 profiles will still produce subprime surfacing  if you use the same rail location.  You'll only get a great loft result with this setup if you either leave the rails off, or place more at the ends of the fillets, which would tricky.  (I'm guessing at this point you don't believe me, so I would invite you to do some experimentation.  You'll see what I mean)

So i'll reiterate, the basic axiom in modeling is "less is more".  As applied to surfacing that means use the simplest tool that will achieve the desired result.  Will an extrude achieve the right shape?  if yes, use that.  if no, then will a sweep or revolve achieve the desired form?  if yes, use that.  if not, use a loft.  Loft is always the last tool of choice, but frequently the only one that will do the job. 

---

Hi @laughingcreek 

Thanks so much for your reply. I absolutely do believe you, as I said I am no expert in this field so I am truly grateful for your correction and I should have composed the post as a question rather than an assertion, I sincerely apologise for this miscalculation, English is not my first language.

What I wanted to say is that it is my assumption that the surface created between (but not including) the fillets would be class A, however I could be wrong about that. I was assuming that that surface, as the result of two perfect arcs would have G2 or better continuity since there is nothing to break that continuity. I wondered if this assumption is correct and  if so does that comprise a class A surface, or have I understood the meaning of class A surfaces (as G2 or better) incorrectly?

Thanks again for your time. I really appreciate it.

[beresfordromeo]

Hi @Hend-Eng 

Thanks for taking the time to read my posts. I should stress that I am not a professional in this field so anything I write is merely suggestive of an approach that I might take. I would always defer to the more experienced users and professionals on this forum for an understanding of best practice but it is always my hope that my contribution, flawed as it may be, will give you a framework with which to explore alternatives.

Thanks again for your time and good luck with your project.

[laughingcreek]

@Hend-Eng - really, this has become more of an interesting back and forth, and less about helping you.  These things can be fun to discuss.  Sorry for the extended conversation on your post.  If you'll indulge us just a bit more.

What your saying about modeling half or a quarter of an object and mirroring is a good approach to know and use.  And it works for this situation.  Thinking this way can also help develop your skill in breaking down large problems into smaller ones.  I've heard the approach your describing recently refereed to as "horizon" modeling, where the "horizon" is the line where a surface boundary would naturally be tangent on either side of a plane.  Turns out this condition can exist in some non-symmetrical situations also, and can simplify modeling when the condition is recognized. 

---

@beresfordromeo wrote ... it is my assumption that the surface created between (but not including) the fillets would be class A, however I could be wrong about that. I was assuming that that surface, as the result of two perfect arcs would have G2 or better continuity since there is nothing to break that continuity. I wondered if this assumption is correct and  if so does that comprise a class A surface, or have I understood the meaning of class A surfaces (as G2 or better) incorrectly?

---

I'm not clear where in the process your talking about here.

But I will say that the surface of the initial loft was where the issues start.   Because of the way it is defines it is a bit wiggly (technical term).  you can cut into it and stitch a surface to it with a G2 condition, but that won't make it a "class A" surface.

lofting 2 "perfect arcs" that are tangent (G1) will result in a G1 edge condition.

The term"class A surface" is a generalized (sometimes overused) term, and refers to more than just the surface edge conditions (G2 etc.)   Ultimately it's referring to the visual quality of a surface when you blow it up to something the size of a car (A is for Automobile), put a shiny paint, on it and look at it in the light.  Fusion isn't really great for creating class A surfaces all the time, because it lacks some of the tools necessary.  It can do it in some specific situations.

But back to the point at hand-

As you progress from sketches, to primary surface, to secondary surfaces, to tertiary surface -  things like curve and surface quality will either stay the same, or degrade.  They will never get better.  That's why it's important to start with the best conditions possible.  Fixing on the back end is frequently not possible, and you end up having going back to the root cause anyway.

To restate what I just said above- a surface will not ever be better than the input sketch, and will probably be just a little worse (another axiom of modeling, closely related to less is more).   Your about to run head long into this situation with a model your currently helping someone else with-

[beresfordromeo]

hi @laughingcreek 

Thanks so much for your comments, I really appreciate them.

On the class A stuff, it seems I have rather misunderstood the meaning of the term or rather lazily accepted the something of the generalized meaning. 

I hope we can come back to this at some stage but it is 5 a.m. here and my insomnia is probably literally slowly killing me so I will try to be disciplined and not ask more at this time.

On the other model , I absolutely know what you mean about this and really appreciate you mentioning it. I was trying to develop an approach for the purpose of teaching the OP (not even sure what this stands for) the importance of other concepts like coincidence, tangency and co-planarity (a truly made up word ha ha). I am hoping that once that is understood and registered I can help them to understand Sketch > Project > Intersection Curve and Sketch > Project > Project to Surface. Once they have understood that I am hoping to move onto the surfacing tools and then eventually onto getting better results splitting faces and bodies etc. Each screencast and description takes an age to write so it is a slow process and not helped by the never ending headache caused by a lack of sleep, I really worry that if I don't keep things simple (at the expense of  surface quality) it will all be too much for both of us.

I have so many question with regards to this though, for instance what happens to the tangency/curvature when you Sketch > Modify > Break a Control Point Spline  vs a Fit Point Spline or the continutiy of a surface when you use Surface > Modify > Split Face, so if you are able to help me to better understand some aspects I will be in your debt. I will leave this for now if that is OK and try to get some sleep but if @Hend-Eng does not object perhaps we could pick this up again in this thread as you suggest.

Thankyou thankyou and thankyou for your valuable time. I will try to take up as little of it as possible.