Continue to Site

Welcome to MCAD Central

Join our MCAD Central community forums, the largest resource for MCAD (Mechanical Computer-Aided Design) professionals, including files, forums, jobs, articles, calendar, and more.

Help with three sided surfaces

I agree with puppet.

It could be done with less feats count.
However the easier way means - not an order im mesh grid. Sad. I play with this topic for some days and I had not found a better way apart of this extra feats to create 4 part boundary. Thanks Bart for models from DE. A lot could be learned from them.
 
Roto, can you post the file as a WF2.0?? Check the model in mesh mode or wire frame...see if you get a nice smooth surface, I suspect you will get a pinched mesh??
 
Design - Engine,


That seems like a far easier way to construct the surface, will try my luck this afternoon. Hmmm I'm wondering what the reflections will look like though...we shall see I guess.


Cheers though, very much appreciated
 
Design Engine,





Just tried your method. You can see that this method suffers the same fate as every other that I have tried. You still get surface deformations where the two COS lins are used to create a four side boundry curves. Whether of not this would transfer onto an injected moulded part or not I'm not sure - what do you think?


Oneadvanatge this method offers is it ease and simplicity....but still dosn't offer a completely smooth solution.


I'm very aware that the model I am using to demonstrate this on could have been achieved by using a revolve & a cut...was used just for testing this method.


Ok $100 *Aus for the first person that comes up with a total solution
smiley1.gif



View attachment 3899
 
if it used CoS. one would conclude ur using isdx.
then just click on the "arrows" which are leading into ur new surface u want the the arrow that looks like this (g2) this will produce ur

=====> g2

--------> g1

-------- g0

Also make sure u have joined ur surfaces.


Edited by: puppet
 
I have only used ISDX to create the COS lines on the surface my 3 sidedboundry surfaceis tangent to. Boundry curves / surfacesare alltagnet.
 
I have some problem using Bart approach. Well, the fact to use 4 part boundary is obvious to me, but ... take a look on pictures:

1) main surf

View attachment 3906

2) front surf curves - I named it "Bart approach", with supporting curves behind the curve from 1st surf

View attachment 3907

3)front surf - with tangency in all needed places

View attachment 3908

and mesh

View attachment 3909

well it is not ideal result but could be. But...

View attachment 3910

there is gap, and front surf with the main one could not be merged.

Finally I stayed with this

View attachment 3911

any ideas?
 
78finn said:
Just tried your method. You can see that this method suffers the same fate as every other that I have tried. You still get surface deformations where the two COS lins are used to create a four side boundry curves. Whether of not this would transfer onto an injected moulded part or not I'm not sure - what do you think?


I wouldn't exactly call it deformations but thereare apparently discontinuities between the surfaces. Making thesecond patch curvature continuous yields better results with the zebra striping. From our perspective, this is good enough for our molded parts, but not sure if it would pass muster in other disciplines such as automotive. Guess that's why they use Alias & ICEM not ISDX.


View attachment 3919





View attachment 3920





View attachment 3921


View attachment 3922
Edited by: mgnt8
 
mgnt8: The lower example is the same as the upper example. I have in every case used cutting away the outer material or like the last image above add inner material for making a four part boundary with success. Every time. The trick in getting the zebra stripes to line up correct with g2 continuity is the smoth generation of the COS curves. Those guys must be created w/o internal edit points. ie: two dig points accross the surface so the curve is as simple as possibe.


You make curves more complicated by picking three times to create the Curve on Surface and that makes later in the surface strange wiggles in your zebra stripes.
Edited by: design-engine
 
That last image is a curve that I projected onto the part after all the surfaces were built & it's clearly got a break in it. Maybe this doesn't matter.Like I said, I've never seen any problems with our injection molded parts.


Below is what the part looks like when I use two straight lines projected onto the quilt to make that trim. Looks a littlebetter. Regardless of how we split hairs (or zebra stripes), the gaussian analysis shows no distortion at all.


View attachment 3933


View attachment 3934
 
All this stuff with trying to create a truly four-sided surface is fraught with difficulty and doomed to unacceptable results.


It's been quite about six yearssince I did this, but your problem is not unique, and there is a solution which will deliver what you want. I seem to recall that you need a datum point that's coincident with the 'degenerate boundary' (point): you use this as the terminus for one of the directions for your swept blend. I haven't looked into your model, so I don't know if you're actually trying this already.


I think there's even a lesson on this in one of the on-line tutorials from PTC University (Advanced Surfacing?): do you have access?
 
for a good reason. I remember once when a PTC dude (who now works for saladworks) told me with enthusiasm, "we can do 3 part boundaries in ISDX now".

I said "dude". "you really need to take my class!".

3 part boundaries are improper use of surfacing. Bad bad bad. and if someone created a tutorial to show you how to do it that person is novice still.
 
Saladworks....
smiley36.gif



I love Soliworks but even I can laugh.....do we get lettuce with that???


Also, from what I remember about this post and I hope I'm right, I think the initial model had some dodgy building in it.....can't remember exactly what it was but it worsened his problems trying to finish the end.
 

Sponsor

Back
Top