Cannot sweep on tight angles?

[Anonymous]

It seems that Sweep cannot work on "tight" angles? Btw, what means "tight"? How many degrees?

In the self-explaining movie below I have a block as a sweeping profile. When I want to do a sweep on the upper triangle it will not work while on the diamond in center is ok.

If I do fillets on the triangle (see the movie) everything turns ok. Ipt file is attached before doing the fillets with the profile already put on the extruded triangle (see Sketch6).

Movie is here:

https://drive.google.com/file/d/1u4v7xlQSJMVEsCrpCuq4UNRD3X_DGccY/view?usp=sharing

[The_Angry_Elf]

I think part of the issue is you are attempting to sweep a profile that is on the same plane as the path.

Typically the profile needs to be (ideally) perpendicular to the path.

If you take the profile and swing it out to an angle (45 degrees if that's the intersecting angle), it might give you a better result.

Another issue is at times you cannot sweep a solid onto itself. It's along the lines of doing a revolve and the solid is revolving material into itself:

At the angles you're using, I see where that might cause an issue.

You might want to look at different approaches such as doing each leg separately. I think if you adjust how the profile is in relation to the path, you'll get the results you're looking for.

[-niels-]

This is a bit choppy perhaps, but i think it's mostly about avoiding path intersections.

See attached file.

[johnsonshiue]

Hi John,

This is a bug. As long as the self-intersection does not happen within the same segment, it should work. I will work with the project team to see if this is a known issue. As for alternative workflow, for this particular case, you will see if will work if the profile is a full circle.

Many thanks!

[Anonymous]

Hi Johnson,

Thanks a lot for the support!

Indeed it is a bug and I would say a rather nasty one because is confusing and, hence, frustrating. As you know, users (especially the newcomers) don't want Inventor or any other program - they want their job done. Being unable to work because of "something fuzzy" will generate frustration and then they will switch to a competing product.

Rather than having such a behavior is better sometimes to have a much simpler feature.

Speaking about "much simpler feature(s)" and Sweep there is a very special case here (see also the others' answers ITT):

"Typically the profile needs to be (ideally) perpendicular to the path."
- @The_Angry_Elf

Besides that It is difficult (at least for me) to create/find a "plane piercing the path", especially when we have non-rectangular surfaces (the diamond from from the provided sample for ex.) or, worse, when we have an organic shape / BRep.

So I would propose an extension to Sweep:

Proposal Description:

Sweep should take

1. any closed loop

2. any point in this loop (including of loop's circumference)

3. the path

And will sweep the closed loop defined at #1 through the point defined at #2 along the path defined at #3 keeping the loop perpendicular to the path. Advantages:

a. We will get rid of all the manual work involved with planes, sketches & the piercing point in order to "patch" the lack of this feature.

b. Increased flexibility (a change in geometry can break the scaffolding described at #a above. Getting rid of this will allow as to change our geometry more easily)

c. Easier part maintenance

d. More straightforward, intuitive learning - generally a better LX.

Proposal GUI:

Simple. Just add another option to Type:

Path & Insertion Point

Perhaps I should note that this approach will be more intuitive since we can select two things (Profile & Insertion Point) in the same entity.

Implementation notes:

The engine is already there, more or less. From my POV building a temporary sketch perpendicular on path & projecting the profile with the insertion point piercing the path is perfectly fine. Of course, the team is free to choose the details of implementation.

[The_Angry_Elf]

John,

One of the things I was attempting to get across was where the profile to be swept lies in regard to the path it will follow.

On you example, your profile was on one of the very same planes (or so it looks like) as the first path "leg" (identified below by the "star").

I typically, in a case like what we are attempting here, is to have the profile drawn on a plane that is 1/2 the angle it needs to "bend" around (see pic below).

Doing so, usually results in a completed sweep as you see below:

Even on far tighter or odd ball type angles:

[The_Angry_Elf]

Doing such I was even able to get it to work on your example part:

I even tried numerous sizes of the cut away and radius.

No angles changed, no fillets at the corners needed.

Am I missing something?

[Anonymous]

Hi Jim,

Thanks a lot for the response!

Nice trick and somewhat logic.

...however this isn't the way in which the things should work IMHO. I want my part to have a certain profile which I've measured.

So, in order to obtain that profile I need to „elongate” the geometry on the bisector in order to have the swept profile projected on the normal (perpendicular) plane as expected.

Since the angle with the bisector can have different values and the profile can have different shapes (including arcs and splines), our trigonometry problem becomes very tough.

[The_Angry_Elf]

John, understood but I think the key point here is to have the profile being swept not to be on the same plane as the path.

Basically the way you have it, the profile is planar to the first path, meaning when it is swept, it's at a zero depth (if that makes sense).

Putting the profile at an angle, any angle other than zero or planar to the path (especially the first), should solve the issue as far as I've been able to test.

I just typically split the difference out of ease, but that doesn't always allow itself to be the case.

According to Adesk instructionals, it just states "A path must intersect sketch plane of profile". Every example I can find all show the profile either perpendicular to the path (even a closed path) or at some angle, never on the same plane. Just a thought.

[Anonymous]

I see what you're saying, Jim.

However if you look again at my movie you'll see that the sweep works pretty ok on two from three paths. The paths on which the sweep works are: the one which my profile is at zero degrees and the one at which is perpendicular - exactly the cases which you raise as suspect.

It doesn't work exactly in case in which you say it should - the 3rd (farthest) path which is at certain angle from the plane of my block.

This is confusing, to say at least, and that's why I consider together with @johnsonshiue a bug.

[The_Angry_Elf]

John,

I agree. I see that in the movie, just wondering if it's the reason the last path fails tho. I know it may not make sense if it works on the first path where it's planar, but I've seen odder things happen.

Besides, what's Johnson know, it's not like he works at Adesk...oh...wait...ummm...;)

But seriously, Johnson will get to the bottom of it I'm sure, he's a rock star here.

[Anonymous]

Thanks Jim for the consolation!

These are Cargo Cult Science for us. Yep, there are enough oddities - just to name a few famous ones:

- You can bake your laptop or graphics card in oven or you can put you Hard-Disk in a freezer

- Of course, there was a time when you should drop your computer on concrete (hard enough but not too hard!)

- The ice on cake was that If you move your mouse continually the SQL query may not fail. Do not stop moving Good old Excel. It seems that they fixed it tho, - now if you MOVE your mouse it stops calculating (!).

We live amazing times.

...and I am guilty too, like you & others. 🙂

Many times I used "TCC" (To Calm the Compiler) or "TFO" (To Fool the Optimizer) "features" so I quite understand you and Johnson's need to fix it.

My experience is that Cargo Cult creates to user an acute level of discomfort and lack of trust (especially if we have to deal with a new user) and hence we will have a high bounce rate to competing products.

Especially today, when we live in such a superficial civilization driven mainly by speed, nobody waits for you to tell them that "well, is better to add 0 which does nothing in order to do a natural scan on your database rather than use an index" when they see a SQL query like: Select foo from bar where baz=blah+0 order by zing+0 . They just bail out to a more "trusty" product.

I hope that I enjoyed you at least a little with my oddities. 🙂

[johnsonshiue]

Hi Guys,

You guys gave me a lot of credits that I don't deserve. I just happen to be on the forum quite often. Without the Inventor engineering team backing me up, I can only post nonsensical information, which I do from time to time. This is team work. I am just a messenger.

This case is indeed a bug, particularly when a full circle works but a pie does not. This does not make sense. On the other hand, I don't know when or how the issue will be resolved though.

John,

Regarding your suggestion to Sweep command, I think it is better to post it on Ideas Forum or start a thread on Inventor Beta site. This general forum is more about issues, bugs, and how-to-do-something.

Many thanks!

[Anonymous]

The idea is here:

https://forums.autodesk.com/t5/inventor-ideas/streamling-amp-enhancing-sweep-automatic-perpendicular...

Johnson,

Of course that you don't deserve credits!

That's why I am here. The people fell not because they are weak but because they are way too strong for too long.

Just letting you know. 🙂