Something missing?

This may be another one of those "How do I approacht this?" issues.

In doing some assembly work with SW one thing that irritated me was just how
easy it is to break mates when editing parts. It would be quite useful to
have the ability to have more absolute mate references (or truly smart
ones).

One example of this is a part that you'd want to mate with another by always
using the highest surface in the Y direction. I don't think that there's a
way to embed this kind of intelligence into parts at the moment. Having
this degree of intelligence would allow you to edit and experiment with a
part and not destroy a million mates in an assembly.

I haven't looked into the internals of Mates, however, on a superficial
first-inspection they seem to be tied to the names of features, edges,
points or planes rather than the feature itself. There seems to be little
intelligence. Which means that, if you were to model exactly the same part
without a single geometric difference except for the modeling sequence, a
"Replace Component" would result in lots of errors.

It is far too easy to blow-up a complex assembly by making an innocent
change in a part. The more I use it the more I understand why in-context
relations and features regarded as problems. The sad part is that such
abilities as defining in-context geometry is one of the big reasons many of
use have looked towards the use of programs such as SW.

-Martin

 

Martin,

You are absolutely correct, creating in context relationships between parts
is easy, too easy. It takes a fair bit of experience and forthought to make
parts and assemblies that work well and retain in context relationships but
IT CAN BE DONE, and too good effect There is a defnite heirachy of suitable
mating geometry. Model vertices are the least desirable geometry for mating,
followed by model edged, then faces, the reason being is that these geometry
types are easily deleted or supressed. Adding a fillet to a part can
"delete" several vertices / faces in the context of an assembly. Using
planes or axis as primary mating picks can help however planes and or axis
that rely on other model geometry for their existence are just as
vulnerable.

Layout assemblies or "Skeleton" parts / assemblies can be used to great
effect when doing this kind of modelling, Greg Jankowski of SolidWorks has
an excellent presentation covering this topic, your VAR should be able to
provide you with this.

Mates are specific to the assembly in which they are created, if you replace
one part with another in an assembly then the mates will inevitably be
broken, however SolidWorks does a great job of helping you fix the
"dangling" mates with a fairly simple UI.

 

Look into Ed Eaton's Trees of Blood presentation which can be found
elsewhere on the newsgroup.

 

"One example of this is a part that you'd want to mate with another by
always
using the highest surface in the Y direction. I don't think that there's a
way to embed this kind of intelligence into parts at the moment. Having
this degree of intelligence would allow you to edit and experiment with a
part and not destroy a million mates in an assembly"


For this specific example you could try:
-extruding a body from the xz plane "through all" in the desired y direction
without "merge results" checked. Maintain this as your last feature by
re-ordering as required.
-mate to the surface of this place-holding part, then hide it
-prior to sw2004 or if otherwise desired, the same could be accomplished
with surface bodies

This would only work well if both mating entities were planer. If there
were not, the same general process could be used with reference geometry
like planes and sketches with pierce-points, but the exact method would have
to be examined on a case-by-case basis.