Radius propagation

[hradford5]

I have a part that I want to put a couple of fillets on, but GM is propagating the fillets onto adjacent edges, that I don't want the fillet on. I have the tangent propagate box unchecked. I'm attaching the model, the edges I want my .156" radius fillet on are edges 6 & 12, the edges that I don't want it on are edges 7 & 29.

 

[Lew_Merrick]

Zeke,

Are you talking about the Fillet 7 (5/32 inch fillet) feature? On my (V17) system it comes in as do not propagate -- and that is what happens. ???

 

[hradford5]

Zeke,

Are you talking about the Fillet 7 (5/32 inch fillet) feature? On my (V17) system it comes in as do not propagate -- and that is what happens. ???

I'm gonna have to get this updated.

 

[Dave H]

Is this what you want to do?

GD, V16 Latest build.

 

[TylerDurden]

Is this in the ballpark?

 

[hradford5]

Is this what you want to do?

GD, V16 Latest build.

Thanks Dave. Upgrading made my problem go away.

Zeke

 

[JST]

I am using this, which I downloaded only a few weeks ago:

PRODUCTVERSION 16,0,2,16496 32-bit

It definitely does the propagate problem. I have a part that it propagates the fillet on so that it runs off the tab I WANT to fillet, jumps a gap, and cuts off a big chunk of the part.

Now, I am using it with XP until I can sort out what I need to do to swap to 64bit win7. No doubt they would blame it all on XP......

Here's a picture of the part. I want to fillet the corner on the flap in the opening on far side. Note it did the one on the nearside of the opening correctly



When I do the fillet, WITH PROPAGATE TURNED OFF, this is what I get (same view, after fillet)

 

[HaroldL]

JST,
It appears that you're simulating a sheet metal part. Try enlarging the "bend relief" in at the end of the flange deep enough to break the surface that the fillet is blending. It should at least reach to the tangent of the fillet (thickness + fillet radius). That should stop the fillet from propagating down the edge of the part.

It looks like the fillet worked on the opposite edge but it did propagate over the relief and filleted the surface. If this part is ever converted to sheet metal that may have an effect on the conversion.

 

[JST]

Harold, that's a fine "workaround".

But it does not address the issue.

What is wanted is a way of controlling the extent of the radius, a way of propagating it only a certain distance along the edge, without needing to produce a large cutout there. This has real-world uses.

It is not unknown, for instance, to punch and bend a tab in the middle of a sheet somewhere, "directly", without making elaborate cutouts and reliefs around it. The tool punches 3 sides, and "wipes down" the tab that is so formed. Naturally there is a bend radius formed there, but it may extend up to or into the flat.

While the "technically right" way is to provide side cuts, etc, the practical real-world way it is done is often as described. The program should handle it. You need only look inside many computer cases to find examples.

Possibly providing a cut of 0.001 might be sufficient, but your sheetmetal house may look at you funny when they find it.

I believe GD would be unable to handle that feature unless the extent of the fillet was controllable, or unless some other more complex and cumbersome method such as the artificial 0.001 cut were used.

Then also, an edge of a machined part may need to be filleted only up to a point. It looks as if one would either do it with a sketch and extrusion, or by very good planning of the order of operations in GM. Or one could ask for a fillet "from here down 35.7mm" or the like. A CNC mult-axis mill would have no trouble making the part.

 

[HaroldL]

JST,
I like your idea of having a distance control on the fillet tool, but I don't know that I would call my method a "work around" it's just the way GD/Alibre works, or doesn't work. I recently turned in an issue to Support that I had with the Loft tool. I asked them why it couldn't function like SolidWorks with regard to controlling the shape from the first sketch to the second. I was informed that GD uses a different core than SW and therefore required a different method of controlling the loft. In my case it needed to have the same number of segments in both sketches to get the result I wanted.

In your example, using a solid to simulate a sheet metal part, it will need different methods than "normal" to get the results you want.
I know of the examples you speak of, tabs punched into the face of a sheet metal part that don't have a bend relief. I've made parts like that myself. (In my day job I use SolidWorks so some of those "tabs" are made with the Forming Tool, something I wish GD would implement). I have designed flanges with and without bend reliefs too.

As for the outside fillet on a machined part, the method you describe has been brought up here in the forum before. At the end of the fillet, or extrude cut, a revolve cut needs to be added to indicate the cutter radius at the end of the fillet. That's just the way it works, for now. I wouldn't expect that turning off the Tangent Propagate option would work in your part either, there is nothing tangent to the edge your are filleting. And fillet tangencies have their own problems, as you well know.

Just for the heck of it I added this image of a sheet meta part with tabs and flanges. Some are sheet metal features, some are part modeling features.

BTW, Have you submitted an enhancement request to Support for a distance control for the fillet tool?

 

[JST]

JST,
I like your idea of having a distance control on the fillet tool, but I don't know that I would call my method a "work around" it's just the way GD/Alibre works, or doesn't work. ......
BTW, Have you submitted an enhancement request to Support for a distance control for the fillet tool?

"Workaround" pretty much means a way to do indirectly what should be a direct function, or is expected to work a certain way, but actually either isn't present or works in some unexpected way. Or, of course, it can be a way to avoid a problem that occurs due to deficiencies in software etc. Your choice how to define it here.

I suppose my issue is that I would not expect the radius to extend BECAUSE THE *EDGE* DOES NOT. It's a question of what controls the fillet.... I would naturally expect the EDGE to control it, since that is what you select to run the fillet along. I would forget all about the "depth of fillet" idea IF the edge controlled the fillet. In reality, the fillet evidently MAY go anywhere that is "touched by" the fillet, as long as at least one surface is contiguous and tangent to the fillet.

I'm not clear on how the various items were all done in your example, but I do see at least one extrude cut, which obviously works.

My EXPECTATION is that if I cut a rectangular hole, and fillet ONE edge with no "propagation", ONLY that edge should fillet, and ONLY for the length of the edge.

When tried, It worked three times as expected. But it does not ALWAYS work that way, apparently, and there is no clear reason why there is a difference. Here I made holes in the piece in three different ways, and put tabs on them. ALL filleted correctly.

THEN I made one with a slight relief on it. THAT ONE FAILED DRAMATICALLY.

Why the difference?

The workaround is presumably to make the relief cut AFTER the tab and fillet are done. But that seems wrong to have to do.






I hear the issue of "that's just how it works". But of course, that has nothing to do with how people work, or want to work..... if SWX, or perhaps several other popular programs work a certain way, and GD requires a much different way, GD may be "wrong", even if the GD way is "better". Windows isn't really very good as an OS, at least historically, but it is dominant, and so everyone just uses it. If one wrote software that contravened all sorts of windows default standards for how the mouse and keyboard shortcuts work, etc, market success should not be expected.

As for submitting a request, I have given up doing any of that until I am on to V17 and full "Design". At the moment, anything I give GD will be dismissed, since I am running V16 on XP, which is not supported, but by my experience and many reports, works fine, and should have no reason why a weird thing like this would occur from that alone. This is only until I can resolve ways to handle some very expensive legacy software that I use regularly and MUST still exist when this machine goes to Win 7. The S/W is much more expensive to upgrade than GD, but GD is forcing the upgrade on me now.

 

[JST]

here's a part that shows how feature order and details affect propagation.

Leftmost: the relief (a partial relief) was put in before fillet. radius propagated through the relief and the material face

middle: relief was put in after fillet. Radius did not propagate through the material face, but two operations are needed to clean it up

right: no relief was used. radius did not propagate through the material face

 

[JST]

I got an answer from support.... I did send it to them.

I STATED that I was alerting them, and TOLD them the same thing that they turned around and told me...... They basically appeared to say the program just works that way, tough cookies, .....appearing to suggest that the part be changed to suit the program..... or that I should use sheet metal (maybe it's not a sheet metal part) which is very far from the way to handle this sort of issue.

Now, if they said "Yeah, it doesn't handle that, and it should, here's the workaround", I'd be fine with that.

It's what I expected, issue was not understood, was dismissed, likely not passed along

 

[Kyle]

JST,

What is the support ticket#?

 

[JST]

JST,

What is the support ticket#?

OOPS, I deleted the email, since it wasn't helpful. I saved some text, though, which I had been thinking of posting, but had decided not to.

What I sent was:

This is in V16, specifically, I do not know about V17/2015 yet.
Problem: Fillets propagate past their proper limit in certain cases.
Conditions noted: fillets on tabs in a sheet when a partial relief exists (see part attached) and "propagate" is turned OFF.
The fillet propagates past the expected endpoint and cuts off any portion of the part extending beyond the line of the fillet.
I would expect the fillet to end where the edge ends. And usually it does. But if there is a notch that extends into the filleted area, but not clear through it, the fillet may propagate past the relief and far beyond the end of the edge being filleted.
I attach a part with three filleted tabs.
The rightmost tab on part has no relief, and fillets properly.
The middle one has a relief, but it was added after the fillet. Note that it takes two cuts to clean up the fillet at the relief, but only one cut was made. It filleted properly.
The left had the relief cut first, and then was filleted. The fillet failed by extending farther than the edge being filleted, cutting off a portion of the part.

The reply:

There is caused when the fillet radius is greater than an adjacent edge:
In this file it is caused by the extrude cut you have added in the file. This occurs because the part is a 3D part. While his behavior may not be desirable it is not surprising. The filet will attempt to propagate at the radius that you have inputted since the curvature will cut some of the material that extends past the extrude cut it will continue cutting material away.
You can see this demonstrated if you move the extrude cut into the part a distance so the fillet will no longer propagate through the rest of the surface.
It appears this part is intended to be a sheet metal part. I recommend using the sheet metal workspace. With the product consolidation emails earlier this month the newest version will allow the sheet metal workspace to be activated. You can download the new version here:

See I told them the cut causes it, and they told me "the cut causes it".

I said that the edge should limit the fillet. They ignored this.

They say "it is not surprising", implying that they consider it "normal and to be expected", in other words, this is basically MY fault for including that feature, and the program is doing what it should. (that's goofy)

They suggest using sheet metal for the part instead. But it may not BE sheet metal, so that isn't a solution. Try this block part. at an inch and a quarter thick, it's not sheet, no tab, but the problem still occurs.





I noticed that the problem occurs in this case when the fillet radius equals the part thickness. If it is larger, the fillet fails with an inconsistency error, and if smaller, it may work OK. With the original part, the fillet was twice the part thickness. There may be an integer relationship issue here.

The inconsistency error seems to only occur if the radius is larger AND there is a cut that is short of the radius in depth at the side. Without the cut, it proceeds OK, and if the cut is deeper than the radius, it proceeds OK. That may not be absolutely true however.

My claim is that the fillet should stop at the end of the edge that is selected, AS IT IS BEFORE FILLETING.

It looks like the algorithm is looking for the edge geometry to be still there as the fillet progresses, instead of noting the edge length before starting, and stopping the fillet there.

 

[Kyle]

As there is another thread started regarding this issue. I'll post my response here as well regarding the fillet incorrectly propagating in the scenario presented from JST. viewtopic.php?f=10&t=18418

This is not how it is intended to work and the fillet should not propagate past the cut that you've made in these examples. The issue here is with ACIS itself and we are currently investigating the problem with ACIS.