Hello Everyone,

Fillet calculation fails for the edges highlighted in the screenshot. What would cause this?

Hello,

You have an extreme amount of points in your construction and your model is out of square. This can cause problems to filleting operations.

Thank you OSTexo. That is interesting because other edges are filleting fine. Is there recommendation on when to do filleting operations? For example, is it better for me to extrude from a square with a taper and then fillet the edges I want, and then move on to add other details?

Hello Baris,

I prefer to do some preplanning on the model and save my construction curve sets so if a filleting or blending operation fails I can back up and try another technique. The first issue I'd clean up is the large amount of control points you have on straight lines. I reworked your model and was able to get a consistent fillet on all surfaces if that is your objective. If you need different sizes of fillets on different areas you can also get there by doing your model planning and doing prefilleting on curve sets prior to extrusion and lofts.

Do you want to get a consistent fillet to break all of the sharp edges of the model? If so what are the desired parameters of the fillet? Unfortunately I'm out of space for attachments so I'd have to make a video to show the construction if that is OK.

How do I go about reducing my control points? I started with a square curve which I extruded with a taper. I then extruded the bottom curve straight down and merged. I was going for a 0.01 distance fillet on all edges including the 5 array of rectangles I booled out from 4 sides of the main shape.

Thank you so much for your help. I'm new at this and I'm trying to learn everything I can do to get better and more efficient.

Hi Baris, so a big complication there is your edges are running into other spots where you've already created fillets previously, like here:





There's also a similar situation at the bottom as well.

When you have fillets that are going to run into each other it's much better to do all the edges for them at the same time instead of in different passes. When you do them in pieces the edges from the previous ones are now part of your object and the new fillets will have to try and deal with them as well. When you do them all together the filleter is better able to know how to stop fillets that are running into each other and build a corner juncture patch where they meet up.

Also doing ones that run into each other in bits makes for more complicated selection - if you have a sequence of edges that meet up smoothly, you can't just select a single edge from that sequence and try to fillet it because that would cause a hole to open up in your model. Here I've made an example with an enlarged fillet to help illustrate:





So if just that edge is selected and not the additional edge that's been created above it from previous filleting, you're basically asking the filleter to do something like this which opens up a hole:



The filleter doesn't like to do that. You'd need to also select the edge from the previous fillet so it could try to run a fillet all the way to the top without opening up a hole. But additionally if this new fillet is of the same radius as the previous one, the little fillet piece that will run to the top can be difficult to create, it's trying to do it as a fillet surface built between 2 previous fillets. And it can cause particular difficulty if the fillets are very small.

It's much better to do all the fillets of the same radius that run into each other all at once instead of separately, that will allow it to make a corner juncture patch between them like this:


Having a corner juncture patch usually makes things much better for the filleting engine, it knows that it can stop the fillet surfaces before they run into each other and it also knows that it can be constructed using a piece of a sphere. If it tries to create fillet surfaces going between existing fillet surfaces of the same radius, with the whole thing collapsing down to a point, that's a lot more complications.

Hope this makes sense - on those notched areas you have it's ok to do one of those separately as long as you do the entire notch, but if you're doing the longer vertical edges you don't want to have done the top or bottom separately, do them all at the same time. And if you have different radius values being used you usually want to put in the largest radius ones first and the smaller ones last.

- Michael

Michael, thank you for taking the time to explain what I'm doing wrong here. I think I selected horizontal loops (like the one you red circled in the first image) and filleted them first and moved on to vertical edges next. Now I know that would cause the fillet engine to throw wtfyouidiot() exceptions. >.< I'll pay more attention to my fillet groupings from now on!

Thank you and OSTexo for helping me with this. Back to MoI fun now.

Hey Baris,
Here's a quick fillet workflow. The thing to note is I do the larger fillets first, then work down to smaller ones. Your original model had some tiny fillets already at the top and bottom..



[EDIT] While I was making the video you made some posts that talked about sizing, and Michael also discussed the "previous fillets" issue too....

Anyway...

ps. At the end of the video I realized I used a "chamfer" on some of the top edges. For the attached model, I backed up and filleted those.

BurrMan, thank you! I learned a ton watching that. Pretty cool to see how the same result can be achieved in less and shorter steps than mine. I appreciate it very much.

Hello,

Pyramid video tutorial is up:

https://vimeo.com/219962551

Hope this helps.

OSTexo

Thanks for the very clear tutorial ... nice clean approach and result.
cheers,
eric

OSTexo Thank you so much this is really cool i was have same problem with curve shape but you fix it with this thank you so much.

Righteous video!

OSTexo, thank you so much. This community is so frickin helpful. I learned heaps from yourself, BurrMan and Michael last couple of days. I very much appreciate all your help guys.

Hi BurrMan,
you wrote "The thing to note is I do the larger fillets first, then work down to smaller ones..."
In my case its work only in opposite direction. In many case, I don't understand the fillet function.
I think we can not generalized it.

Fillet1 works from small to lanrger fillets.


fillet2 try to make fillet from large to small. The small fillets are "corrupt" (not sure with my english ;) )


Scene file

Another try to make 2 fillets on to the same object. One edge can't fillet with 0.1. the other edge works fine with 0.1
the edges are clean and only one segment.



Cheers
Jörg

Hi Jorg,
""""""Generalizing?""""""""""""

Yes! It's a general rule as A small fillet will fit into a larger area, easier than a large fillet will adapt to a small area... It's just the way it works. But that doesn't mean a scenario cant be created where we can get the fillet engine stuck.

Here's a video where I do both your files. The .1 fillet WILL go!



However,

"""""""" the edges are clean and only one segment""""""""""""

Well, as you can see on your second file, I show that your object is not exactly a planar cut and squared up to the arc that is defined. This creates a very small "angle" between that face and the arcing surface around the outside. Not quite a "simple little fillet"!

Then you are working in "Meters" which is a larger value system, but typing in numbers like ".1" etc... When you make the numbers more disparate like that, then it's easier to run into problems with calcs going on behind the scenes.

With that said, it's noted with MoI that we ALL would like a better fillet library. We're hoping to get one in the future.

With your sample though, going through and trying to do "21 different values and 21 different edges" will most likely lead you to cramped quarters and needing to move out to a different fillet engine. Plan ahead.

I keep ViaCad on my system JUST to fillet things I get stuck in MoI!