Inconsistent Fillet Behavior

[Anonymous]

In the screencast below you can see that I have a sketch of some "stairs" that intersect with end "posts". After making a 3D body and applying a fillet to the "posts" to make them round, I expect that the steps will follow the new curves and not remain square. This only happens on one of the three new curves. I've applied all the sketch contraints that I can think to but nothing helps. Please advise what I can do to make the "steps" also follow the top curves.

Here is the F360 model.

Thanks!

Ian

[etfrench]

The walls are in different positions relative to the steps.  The right side wall is flush (horizontally) with the bottom step and the left is not.   The fillets look correct.

p.s. The linked file appears to be for some other model.  It would be better if you attached the actual file (Use Export|Archive file to generate the .f3d file).

[Anonymous]

@etfrenchsorry, but I don't understand what you're suggesting. Please elaborate on what you are suggesting I should do to fix this. The line in the sketch from the top of the "posts" and the top "step" are all the same line as you can see here. I don't know how to make them any more in the same position.

Regarding the files... I'm not sure what's up there. That is the same file but it apparently is showing an old version instead of the most recent version. My Fusion 360 has no option to export as .f3d file. There is only .step, .igs, etc. However, sharing in this way has always been sufficient in the past.

Thanks!

Ian

[etfrench]

It should be the top option in the export dialog:

The fillets at the top step look identical to me.  The walls are different at the bottom step, so the fillets on that step are only on the right.   Perhaps a screenshot with what you think it should look like marked in red would help.

[etfrench]

Without your file, I can only guess.  Did you create all of the fillets in one command or did you do them individually?

[Anonymous]

I created all the fillets in one command. If you go to the shared file link in my original post, there is a button there to download the .f3d file. It appears to be displaying the complete model now.

In your screencast you correctly duplicated the part of my model that works. Would you try adding the line across the top where my issue exists and see if you get the same issue?

Thanks!

Ian

[etfrench]

The A260 files are a pain to deal with.  You have to wait for an email with the download link, then you have to upload the file.  If you attach an exported .f3d (Archive) file, it will be saved in your download folder and can be opened directly.

The fillet appears to behave differently depending on how you create the steps.  I've attached my file so you can see how it was created and how it behaves differently.

[Anonymous]

It seems that I can't export .f3d files because my file has another F360 model embedded in it...

To properly duplicate my problem, you need to change your model as shown below. It seems that your file has the same problem. Or is this an expected behavior? Personally, I would expect it to wrap the curve on the top step the same way it does on the bottom step.

Thanks!

Ian

[daniel_lyall]

The fillets in your posted link are identical it may just be a graphics error and they are derived from a slightly different element 

[Anonymous]

---

@daniel_lyall wrote:

The fillets in your posted link are identical it may just be a graphics error and they are derived from a slightly different element 

---

So, a bug then? Any ideas for a work around?

[daniel_lyall]

No not a bug take the 2 fillets out from the right side a do them by there self show the one from the top level  you get this 

then the second one you get this, It's attached to a different shape to the others the two on the left have both a same level bottom the first one on the right is from the same level the second one on the right is on different level's you measure them they are all 1.5 

Now just for fun I split body at the bottom faces and you get this 

Then adding the fillet back in as a fillet by it self for the bit on the step you get this

As you can see it's to the size you set and looks the same

[etfrench]

I can't see any difference between the left and right side top step fillet.  The bottom step fillet on the right side extends into the second step whereas the top step fillet on the right side does not.  The behaviour is entirely consistent.

[daniel_lyall]

Where the line is sitting on that last pick is the center of the wall so a 1.5 fillet that is where it will land as half the wall is 1.5, it's a not seeing it correctly if you measure it, it's bang on.

to me it looks slightly wrong but measures correctly for what you have set the bottom step is just over 1 mm change it to 1.5 and redo the fillet it looks correct

[Anonymous]

There's definitely something weird going on with the fillet. Check out this simplified test I did. Depending on the order of events I can get three different behaviors from fillet. The second (step #3 below) one I'm not even sure how to get F360 to produce it in a way in which it will actually stay in place. In the screencast you'll see me do the following...

1. Create fillet with steps and no back wall. This creates the desired behavior of the step wrapping around the fillet.

2. Adding the back wall after creating the fillet. Retains the desired behavior.

3. Dragging the fillet step to occur after the back wall. Creates a rounding of the step inwards.

4. Open the fillet step to edit it but not actually changing anything. Creates the undesired squared step from the original post.

[etfrench]

Each time you make a change, the model is re-computed.  The model was consistent after each change you made.

[daniel_lyall]

can you post that file @etfrench that is what I am see aswell, I just wont to see if recompute regens it to what it should look like

[jeff_strater]

Yes, this is certainly surprising, @Anonymous.  I kind of understand what is going on.  The result you get after editing the fillet is the result you would get if you created the features in that order (that is, the fillet last).  So, technically, that is the correct result.  Editing a feature is kind of like deleting it and re-adding it.  So, the edit shows the correct behavior for that creation order.  

What this case shows, in a pretty subtle way, is that feature reorder is not quite identical to changing history.  Most of the time you get the same results, but in rare occasions you see things like this.  I'll dig into this deeper with the parametric feature guys tomorrow, just because they will want to see this one.  More news if I learn anything earth-shattering.

Thanks for an interesting CAD problem to explore!

Jeff

[Anonymous]

---

@jeff_strater wrote:

Yes, this is certainly surprising, @Anonymous.

Thanks for an interesting CAD problem to explore!

---

Thanks, @jeff_strater. I look forward to hearing the result. While you're at it... I believe there are two correct answers to this problem: 1) step follows the fillet outword and 2) step curves inwards. It would be nice to be able to somehow select which effect is desired. The case where the step just squares up doesn't make sense to me but perhaps there's a case where that would be preferred, also.

Thanks!

Ian

[jeff_strater · Accepted solution]

Hi @Anonymous,

I understand better about what we are seeing in this model.  I'll try to explain it as best I can, without going too deep.  The summary is:  I would probably consider this a bug, but, being honest, I can't promise a fix any time soon.

I've created a simpler model.  Here is what is going on.

At this point in my simple model, I add a fillet to the highlighted edge:

The edge that is selected will be important later.  Then, add the fillet:

Then, add the final extrude:

all of that is exactly what we would expect so far.  Next, we reorder the fillet to come after the last extrude:

This is where it gets interesting.  The fillet result here is because that original edge we picked for the fillet has been split into two edges.  If we roll back before the fillet, you can see the two pieces:

So, this is what produces the concave/convex fillet result, because Fusion fillets both of the split halves of this edge.  Now, when you edit the fillet, Fusion only picks the top half of the edge:

This is where the bug is.  For consistency, Edit Feature should select both halves of the split edge.  But, this is why you get a different result after edit - only the top half is filleted.  You asked about options to get both results.  That does exist, in that you can selectively choose to select both edge halves by holding down CTRL (Cmd on Mac) and selecting that other piece:

Hope that made at least some sense.

Jeff

[Anonymous]

Thanks @jeff_strater. That makes complete sense and its not a critical problem for me but I did want to report it and better understand it. Through this thread, I've learned how to sufficiently work around it and get the desired results.

Thanks!

Ian