When to fillet or chamfer

[grooveman]

Hi.

Just a general question: At what point in your design do you apply your chamfers and fillets? Personally, I have found it problematic to apply them too early in the design process... in fact, I typically wait until the very end -- especially when dealing with complex shapes that have a lot of edges. I'm not talking about assemblies here -- typically just any given body. When the body is complete, that is when I apply my fillets and chamfers.

I do this because I find that it makes subsequent revisions difficult. Going in and changing any geometry after you have done this usually breaks things.

But -- I also find that, even when waiting until the end of the design, things still break when I change a fillet or chamfer. No matter how minuscule the change, If I end up with 6 fillets (for example) at the end of my design, if I go in and change one, all subsequent fillets break -- even if they do not reference the same edges or faces. I can only change the very last one and suffer no consequences. It happens 100% of the time, and makes things difficult when making revisions. It is better than applying fillets and chamfers in the middle of the design where critical geometries can break, but it is still a pain because I usually wind up re-filleting/chamfering the whole project over again whenever I make a subtle change to a fillet or chamfer early in the chain. I cannot help thinking I'm doing something wrong here -- something inefficient. Is there a better way?

Thanks!

G

[drmacro]

Nope fillets/chamfers are very prone to TNP, no matter when you do them. It is the nature of fillets/chamfers to depend on generated geometry and that is prime for TNP. (Topological Naming Problem)

If you can be somewhat confident that things up the tree (i.e. before the fillet) won't be changed, then early is good.

And, if you put them at the end, it means breakage is limited to the last few operations. This is somewhat easier to repair, when it breaks.

Any time you can put the fillet or chamfer in the sketch, it a good idea to do so.

Unfortunately, there is no 100% solution.

[Shalmeneser]

100 % with drmacro.
Sometimes, the fillet is needed at a certain step otherwise it's too difficult or impossible later.

[chrisb]

I dare to disagree, at least partially, and only if you follow the advice for creating stable models.

If you change only the fillets themselves, then it doesn't matter, whether you make them early or late: the following fillets will most probably break.

If you change other things too, then it is at least worth considering to do the fillets early. In such models every change in every feature can make the fillets break - and most probably does. So having many of them before the fillets means to have many opportunities to break the fillets. Having only few of them before the fillet means to break the model with less probability.

[GeneFC]

Just a general question: At what point in your design do you apply your chamfers and fillets?

Not quite a direct answer, but I always try to place the "fillets" in the original sketch when possible. This is 100% robust.

We often see examples on the forum where someone creates a square corner in a sketch and then tries to round the corner with a fillet when back in Part Design. Totally the wrong approach.

Of course there are many cases where the fillet cannot be drawn in the sketch. In those cases the advice given above applies.

In other words, "it depends."

Gene

[TheMarkster]

I have a pull request pending that will hopefully prevent TNP issues in those rare cases where all the edges are filleted the same radius. There will be a property called UseAllEdges that when true will fillet all the edges regardless of how many there are. This will work similarly for chamfers, too. This is for Part Design only.