Undercut Detection

Is there anyone out there who uses the "Undercut Detection" in SW05? Has
anyone noticed that it's completely wrong?

Here's a very simple example part: www.dezignstuff.com\swparts
\undercut.SLDPRT (using the default config)

Here's how the undercut analysis should read for this part:

Dir1 undercut: 0
Dir 2 undercut: 0
Occluded undercut: 0
Straddle undercut: 0
No undercut: 17

But instead, it reads this way:

Dir1 undercut: 5
Dir 2 undercut: 5
Occluded undercut: 0
Straddle undercut: 0
No undercut: 7

Looking at it, the part obviously has no undercut faces when the Front
plane is the direction of pull. SW seems to think that the only types of
faces that cannot be undercut are faces with no draft, in other words, ALL
DRAFTED FACES ARE UNDERCUT.

I don't have any problem with the calculations, I just have a real problem
with the classifications. Instead of "Direction 1 undercut", the interface
should read "Pulls from Direction 1". Likewise for Direction 2. What most
people think of as "undercuts" are actually listed under "Occluded
undercut". See "Other" config for occluded undercuts.

Anyway, is there something I'm not getting or has this function always been
simply wrong?

Matt

 

Matt,

It's very hard to understand (and I don't).

To me, an undercut is an area of a feature that will require a slide or some
other type of moveable form,,,that's it...

This thing doesn't seem to give you any usefull information. Or if it does,
it's obfuscated to the point of uselessnes.

Probably looks real slick in a demo though


Regards

Mark

 

Matt:

I think the mistake you are making is to think that when you use a
plane to define the direction that Solidworks interprets it as a
parting plane. If you want to get the results that you are expecting
you need to insert a parting line and use that in the Undercut
Detection. It then gives the results as you predict.

Mike

 

wrote in @g14g2000cwa.googlegroups.com:

The same discussion is going on in the SW subscription forum. It works
differently, but still incorrectly when done with a Parting Line feature.
Anyway, the help describes its use with pull direction and with PL. With
the PL feature, two faces of the undercut side core from the part
"undercut2" are identified as undercuts from the same direction, but the
other 3 faces of the side core are shown as not undercut. This is the
biggest giveaway that the "undercut" concept has not been understood by the
folks who defined the function.

I'm not making the mistake you point out. I understand the difference
between parting plane and pull direction. Just by identifying a pull
direction, SW should be able to tell if a face is fully or partially
obscured by the solid in the direction of pull without identifying the PL.
If a face is not obscurred by the solid in the pull direction, and it is
drafted in the direction of pull, then it is not undercut.

Part of the problem is that SW says that all faces on side B are undercut
when viewed from side A. That's not useful information. Useful
information would be telling me which faces cannot be pulled from side A or
side B.

They should get rid of the "Diretction" part of the undercut face
identification. Either a face can be pulled from A or B or neither, there
is no distinction unless a set of undercut faces are fully surrounded by
side A faces.

Just out of curiousity, what part of the information given do you find
useful?

 

I wonder if there is a difference depending on the service pack? When
I inserted a parting line and ran it I did get the correct results
(SW2005SP1.1).

I don't actually find it particularly useful. I generally use the
draft analysis instead.

 

wrote in @z14g2000cwz.googlegroups.com:

This has been this way since it was implemented. I complained about it to
SW and got the old "this is functioning as designed" routine. Apparently
the people who design the functions are infallible.

Anyway, I'd be interested in what you think "correct results" are.

I love the draft analysis tool, but it doesn't tell you directly where
undercuts are. You can only infer that from seeing red faces in a sea of
green. Unfortunately the undercut tool doesn't give that info either.

matt

 

That's true, but what benefit do you get in distinguishing "Undercut" from
"Direction1 Undercut" unless it allows you to knit surface bodies from the
various types? If you're going to do that, you might want to go the route
of doing the same for the draft as well, so you'd have no-draft faces
grouped into no draft, no draft dir1, no draft dir2.

Better yet, instead of grouping by direction or side, it would be cool if
they could group faces into sets that constitute a single "undercut
feature", so you'd have "Undercut1", "Undercut2" which would help you
construct lifters or slides. Really, the direction or the side of the
undercut doesn't matter.

Matt

 

Just to add a little here, if you ever use Splitworks, it groups faces and
knits surface bodies after doing a draft analysis, and allows you to create
custom groups and move faces between the groups manually. I don't think it
would be too difficult to automate this somewhat so that the sets of
contiguous undercut faces are automatically grouped and knit into a single
surface body.

Anyway, that would be slick and solve some of the difficulty in making
slide/lifter geometry.

Matt

 

"Anyway, I'd be interested in what you think "correct results" are."

What I referenced as correct results are what you indicated in your
first post:

Dir1 undercut: 0
Dir 2 undercut: 0
Occluded undercut: 0
Straddle undercut: 0
No undercut: 17

Except that for No undercut I get 19.

Mike

 

Just noticed the undercut2 part.
You're right it does screw up the side core.

Mike