Variable radius problem

[jcsantilli]

There is a way to have a variable radius in this edge? Can't figure out how. Thanks

[jeff_strater · Accepted solution]

yeah, today you can only apply a variable fillet to one edge at a time.  To be honest, I'm not sure why.  Let me look into that next week - I'm not sure if that is a real limitation or an artificial one.  I was able to get it to work, very painfully, by setting up 6 variable fillets, initially setting their values all to .2, then setting one start and the adjacent end to .4 and another pair to .1.  Model is attached.

[jcsantilli]

Thank you Jeff for taking the time to help me with this. Yes, I think the Fusion team should take a look at this. I don't understand how doesn't have the option for chain selection.

Thanks again

[jcsantilli]

I'm still playing with this file and found out something that I think is not correct.

The part in question is a body and when I applied a variable radius keeps an edge that in my opinion should not be there. Any idea why?

[jeff_strater]

there is definitely something funky with this design that I don't yet understand.  Back to the original problem.  Variable Fillet does allow multiple selections, but the selections must be tangent continuous - still need to check on that.  In this model, that is not the case.  The easy way to check for tangent continuity is to use the Fillet command - Fillet will not let you select an edge where the adjacent faces are tangent.  Yet, in this model there are a few in that cutout.  All the blue selected edges are not tangent edges:

This is what is causing Fillet to not select that whole chain.  So, I looked into the sketch that made that cut out, and sure, enough there are a few problems.  Both of these points are not correct connected to the line and arc on each side, and are not constrained to be tangent.

zooming in, you can see the problem:

so, I thought, "aha, that's the problem, I'll just fix that, and Variable Fillet will work", so I did:

But then there are still problems.  There is still a tangent discontinuity here:

so, I went even farther back, to the sweep.  It is simple - just an ellipse swept along a line/arc/line path.  The path is constrained to be tangent:

but, yet, non-tangent edges are generated:

and that is a mystery to me.  Will have to dig in deeper next week.

Also, I crashed several times working with this design.  Not sure why.  I notice there is a warning on the first sketch, but I assume that is because you cut other stuff out to make the case simpler.

[jeff_strater]

if you are still interested in the underlying cause of this, I did get a response back from the modeling kernel team.  It turns out that Sweep with taper will only guarantee tangent-continuous edges in the result if the profile is exactly perpendicular to the sweep path.  In this case, there is a 73.53 degree angle between the two, and that is why those edges are non-tangent.

interesting problem, I learned something from it...

[jcsantilli]

Thank you Jeff. Good to know what the problem was. I also learned something from you!!

Thanks

[chrisplyler]

Good detective work Jeff. It kinda makes sense if you think about it.

Here is something weird: I swept the ellipse from a perpendicular plane (in this picture I've already cut the body off to match the angle of the original). I get decent zebra around the FIRST tangent intersection, but not around the second one. What gives?

[jeff_strater]

can you share that model here?  I'm not certain what is going on, TBH.

[jcsantilli]

What is that zebra thing. Don't understand what it does.

[chrisplyler]

Jeff, here is my model.

The OP's model was in the main hierarchy, so I did my work in Component2.

[chrisplyler]

It lets you see where a surface's curvature is smoothly continuous or not.