Zebra Stripes - Tangency - Loft

So, there I was sittin' there and minding my own business and then "Swx
is unable to obtain required memory..."

That is not the point of this post. We'll leave that one to Good Ol'
Bob Zee.

I am drawing a fairly complicated little part that has a number of
radii and an intricate loft. This part needs to be completely smooth -
perfect tangencies, etc. How do I go about making sure that it is
correct? Is the 'zebra stripes display' the way to go? If so, how do
I read and understand what I am seeing?

Thanks, group. You guys rock.

Arlan
...driving 90 miles to work - one way...

 

Well, after reading through the help topics. I have mostly answered my
questions. This part is not making the zerbra stripes very friendly,
though...
:~(

 

Zebra stripes are for checking C2 (curvature) continuity. They won't
help much for C1 (tangency).

Look into the "Deviation Analysis" tool to check face angles at edges.

Sometimes you can ask for tangency or normal to plane on lofts but
still not achieve it.

 

The fastest and surest way to check an entire model for c1 tangency is
to turn the display to HLR wireframe, and change the tangent edge
display to 'as phantom'. (under system options-display/selection). All
tangent edges are dashed, all non-tangent edges are solid lines.
In my opinion (I think that's implied, but I have to add that since
aquiring my own Cliff) I find zebra stripes to require way too much
rotation and zooming to catch the edges that are just a little off -
using HLR nails the non-tangent edges right away in a non-ambiguous
way.

The next step was mentioned by the Tick - if you have a hard line
indicating non-tengency, check to see if the non-tangency is big enough
to matter by using deviation analysis. If its a fraction of a degree,
it probably won't matter and will get polished out when making the
tool. Most of the solidworks surfacing tools (loft, fill, etc)
frequently are a little off - the SolidWorks guys have told me they
think anything within a degree is really good.
Be especially careful of certain settings.
- 'Optimise surface' on a tangent surface fill will give you surfaces
that are sometimes OK, and other times are 10-20 degrees off tangnet
(while turning it off gives really nice connections within a fraction
fo a degree).
- I find it iroic that if you use 'curvature continuous' as a start or
end tangency on a loft you can end up with a deviation analysis that
can be 2-10 degrees off tangent!
- Also be careful of edges between sections on a variable radius fillet
- the different fillet sections might not be tangent to each other and
might require some repair with surface fills

Ed

 

You have a "Cliff" of your own now? Does that mean you have achieved
Banquer-hood?

 

Banquer-hood? Dare I dream?
Alas, I must humbly admit that it is beyond me to even attempt to
compare to him, and I regret it if anyone read my post as an attempt to
latch on to his stardom. I am just a man - he is the legend, the
institution.
Ed

 

Ed

When you get a spare month or so please write a book on surfacing with
SolidWorks.

I'd like to put in an advanced order for a copy now.

John Layne
www.solidengineering.co.nz

 

<G>.
I'd wager someone recently confused him again on C2 vs. G2 too.

 

Hey put me on the order list too

Ben

 

so are the new 07 C2 really C2 then or is this another case of within a few
degrees being called good enough by SW?

 

Cliff,

Got that profile yet?
How long exactly should it take you to post a simple 2 axis profile?

Tom

 

I haven't touched them for a while in 07 so I don't know if there has
been improvement. But... well, hell, I don't know what I can say.
By the time I do the next SWx Word it will all be out in the open and I
WILL mention whatever the truth is (even though, oddly, it seems to
bother some folks) because stuff like this is important to know.
I feel I CAN say that there a lot of people a lot smarter than me at
SWx who are seriosuly trying their asses off. But I get the impression
that it is a seriously tough problem and at some point they have to
accept what they can get so they can give us SOMETHING. But I honestly
don't know if they beat it over the last few months or if the new stuff
in '07 will be more of the same. Let's hope, huh?
-Ed

 

Thanks, guys. That's nice to hear
here's the problem with trying to write a book -just an actual example
from work today.

Backstory-
We started the week working on some parts that originated at an outside
ID firm in the middle of last year. We tweaked the design, engineered
it (adding ribs, mounting locations and whatnot) and made prototypes.
Last week we get last minute direction to radically change the ID, and
we had a week to do it.
I start the week interpreting the ID sketches and come up with a new
shape trying to reuse as many of the existing features as possible to
keep down the rebuild errors, especially since the previous owners of
the data used the 'master model' approach.

So, point 1 for the book from this sample - try to keep as many face
and edge ID's the same even if there are radical changes to the entire
shape. How do you do that without going into excessive, boring detail
on how SWx thinks, and how do you produce samples that can take the
student through the whole face/edge ID issue (in other words, create
samples that are bad with lots of lost ID's, then taking them through
the same sample with good Id's?) And when you mention this real issue,
how do you avoid the accusation of 'bashing SWx', even though it is
true and must be dealt with in every model, not just surface models.

Today:
I pass along the new, shelled, ID model, and the engineers start doing
their magic.
They have a nice, stable model, and change one dimension to add some
extra clearance because they want extra tolerance around a mating part.
The model 'blows up'

Point 2 for the book - mutual surface trims are notorious for
forgetting which part to keep and which part to delete - even a simple
dim change can cause the trim to flip manifold sections. All the
errors in the tree come from a single surface trim forgetting what was
supposed to be kept - a part of a surface manifold (insert definition
of manifold here - that will take a while) that was supposed to be kept
was deleted, and a part of the manifold that was supposed to be deleted
was kept. The engineers were perplexed (they don't deal with surfaces
much, and are a little leery of them for just this sort of reason), but
I have multiple hundreds of hours logged dealing with just this issue
so its no big deal to me.
So do you put this in the book because it is common and expected, or do
you leave it out because someone will think you are 'bashing the
software'.
And do you stop to make a larger point about the software - a valid one
at that - that it is always risky to use graphical, on-screen selection
(the only option with a trim - 'as far as I know'-, but one that
is just as risky with 'contour select' when you can instead use the
much stabler 'convert edges' and use a new sketch to create a feature)
knowing that you again risk claims of 'bashing the software' or someone
else, even though you can back up the claim with clear samples of the
risk? And do you stop to back up the claim with samples, or do you
just press on? What important points that can save the reader dozens
or hundreds of hours do you leave out?

Point 3 for the book - the engineers were trying to modify the
selection set for the surface trim, but couldn't. The manifolds were on
the screen, they could see them, but they couldn't pick the sections to
keep what they wanted.
I had to stop them and educate them that you can only chose parts of a
surface trim that are 'shown'. I instructed them that they had to roll
back before the surface trim, show all the effected bodies/manifolds,
then they could roll forward, edit the feature, and correct it. I do
this in a few seconds because I've done it so many times, but it took
them minutes because they weren't expecting it (and I don't think
anyone should expect that)
So how do you handle that situation? Do you set up a sample model with
an error just so they can learn about this stupid idiosyncrasy? Do you
call it what it is - a stupid idiosyncrasy - knowing that folks will
then lambaste you for bashing the software, etc, etc, etc (sorry to
sound like a broken record on that one, but it has come up) And then,
once there is a published book, SWx fixes this oversight and your book
is out of date and you have to spend the next month updating it?

And that is just one part from one project - we haven't talked about
anything more complicated than a basic surface trim.
Would I love to share what I know to prevent someone else form having
to waste their time on the same learning curve? You bet - that's why I
spend several hundred hours (seriously, I track it) each year working
on SWx world. It seems like such a waste for me to have noodled this
stuff out and to keep it to myself.
Would I like someone else to go write a book so I can learn from them?
Absolutely - add me to that list you guys started. I can think of at
least one guy - that humbles me - that I want to write that book.
Do I want to deal with the political and logistical quagmire of trying
to do it myself - no way.
I'll just take a (very) little crap every year after SWx world, thank
you (along with a lot of generous compliments, and thanks a ton for
those - it justifies the hundreds of hours). What gets me about that
last one is that I (so far) never get criticized on veracity of
content, just tone - I must come across as a dick or something. I
sincerely look for issues about content, because I really don't want
to say something that isn't true. However, tone I can't figure out
a way to help -I struggle with that more than anything, and, if at
the end of the day I come across as a dick to my core (and I really try
not to be), I guess I'm a dick. I won't revel in it or exaggerate
it, but if trying to say what's true and verifiable and necessary I
earn a couple of Cliffs, then I guess I have to accept that.

So thanks about the pre-orders on the book, but someone else will have
to cash in on it. I hope they do the subject the justice it warrants.
Barring that, I dig the WIKI idea that was brought up here a couple of
months ago. I would be hapy to contribute to that (though, again, I
will run into serious issues when something is presented that just
ain't true. I have a flaw about being pathalogically honest, and that
seems to cause problems in some circles.)
-Ed

 

On 22 Jun 2006 23:35:55 -0700, wrote:

Ed - your too hard on yourself. As for 'software bashing' - well its
not even in the same ethical league as seal clubbing. so carry on
'software bashing' , because it is only when experts like yourself
point out the shortcomings that software is improved.

On an aside - its over ten years since SW came out and totally
revolutuionized 3d for the great unwashed mass. Ay a hunch is 10years
the lifecycle of a product before it gets too unwealdy and cannot
adapt to new ways, - so what is the latest software that we should
keep an eye on as the replacement for the next 10 years? Who are the
new kids on the block with the radical way of doing things? Or should
we raise some venture capital and write our own, as we have over 10
years experiance of how to and how not to do it ;-)


Jonathan

 

Oh my goodness! That is awesome. That works. It really does. I do
have another little issue, though. How do I do the deviation analysis?
Where do I find this comman/utility/option?
I have only used surfacing for 'playing'. This is my first attempt on
a part that will be used to create tooling. I am more than a little
nervous about it.

Thanks!
Arlan

 

I was expecting to see that C2 would be just another condition option in the
existing features so to see boundary surface was somewhat surprising...
My impression from the little I have seen is that it has been impossible to
integrate C2 into the existing program and they have needed to reinvent loft
especially for it... but I could be wrong about that.
Also I am somewhat disappointed by what seems to be actually rather limited
surface pull functionality which also seems to indicate underlying
difficulties.
Maybe though it is just a half finished tool to be spread over several
releases as per usual...
Improvements in splines and continuity have been a long time coming.
I too really hope they end up being genuinely useful tools.

 

Wait about 18 months. There is something new coming that has some
promise. On the plus side, its a new kernel - that is critical (I
think the BREP is about played out and its time to move onto the next
technology). On the down side, I wish the company had more mechanical
designers consulting on the development. Oh well, at least there's a
chance, eh?
Ed

 

Sounds interesting ! - hurray - and only 18 months to go.

Doesn't sound good - who are they writing the programme for? I just
hope there is lots fo drag and drop from lots of different places and
and lots of simple Wizards - none of this $ string business to get a
simple Design able to work.

Jonathan

 

You have to be in a part, not in an assembly. Tools/Deviation Analysis. Pick
the edge or edges you want to check. Hit Calculate. React with suitable awe
or outrage, depending on how well SolidWorks matched your intent.

Jerry Steiger
Tripod Data Systems
"take the garbage out, dear"

 

Should still be the same one: ParaSolid, right?
Perhaps just a new rev level ...

HUH?

BTW, It's not the kernel that is at issue AFAIK. Bear in
mind that such firms as UG use it too. Don't go all jb on us <g>.

Now I have a question .... HOW are they going to do such things
AND remain a constrained parametric history-based solids modeler?
You'd have to manually apply constraints to all those control points ... ?