Small Holes Continued (On Thousands of Holes)

I dug up a good thread where a guy named iQ answered how to do
thousands of holes in a part. That thread is here:
http://groups.google.com/group/comp...nk=gst&q=pattern+slow&rnum=1#5854c2e2763eb906


I couldn't reply to that thread because it was too old but I have a few
questions for iQ or whoever has gotten around the problem of SW being
slow when trying to pattern many holes. We have a similiar problem
where we are trying to make a large photomask with the same pattern
over thousands and thousands of times. We tried making a part file of
the photomask pattern and then inserting it into a new assembly. Then
we tried doing a linear component pattern of this part in the assembly.
This still took way too long (in fact I gave up and ctrl-alt-del the
mofo). What I don't understand from iQ's post, is that he made the
pattern as an assembly, not as a part. How do you do that?

Thanks
pope

 

good morning pope, iQ here. let me review what i do for this
situation.

i have a model that has contact pads (solid model of an item that is
either a round cylinder, a square with rounded corners , ... all
extruded to .001 thickness.) in various arrays to compose a pattern,
that are arrayed again to fill a quadrant of a round PC board, and
another array to fill all quadrants of the PC board. we model one
contact pad as a part model. we then make an assembly model of the pad
to create the first array (this ends up looking slightly weird because
of the size of these elements). we then take this assembly model into
another assembly model that arrays the first array to fill a quadrant.
and again we take this assembly into another assembly to fill all 4
quadrants. we then take this assembly model into another assembly
model where we add the PC board and the rest of the features either in
this assembly or in the model of the PC board part model. the tree
structure looks like this.

final assembly model (where the final array is assembled with the PC
board, which is what we use for detailing the model on a drawing)
--> 4 quadrant array assembly model
--> 1 quadrant array assembly model
--> first pattern array assembly model
--> contact pad part model

we do this with solid models and i have not tried to do this with
surfaces which you might be using for your photomask. i cannot send
any models due to propriatary product it is depicting. the modeling
methods, i can discuss. i have tied this in a single model with
multiple bodies and as a single body, and as you have seen this is not
a workable solution. simplification of the first part model is
absolutely necessary. have simplified and final configurations in all
files so while we work with the locations of objects we are in
simplified mode and when we get to the drawing we are in final
configuration. helps with the generation of the model and helps with
the detailing process, still not the best but at least it is now able
to detail. iQ

 

just noticed that i did not answer your initial question. an array in
an assembly model is component pattern available from the insert
pull-down. iQ

 

iQ,

Thanks for the detailed reponse. If I am to understand you correctly,
SW will bog down if you do a large pattern off a part file in an
assembly, but does not bog down as much if you insert a single part
into and assembly, and then pattern that assembly many times in a new
assembly. Am I understanding correctly? I just did a test of this,
with a 1 inch square part, extruded to .001". I then inserted this part
into an asembly where it is the only part in the assembly. I then
inserted that assembly into a new assembly and patterned 100 x 100.

I then tried the same action, but patterning the part file in an
assembly. My two methods are below:


"Assembly Method"

--> assembly model of 100 x 100 pattern of the
assembly below
--> assembly model of one part of below
--> 1" square x .001" part
model

"Part Method"

--> assembly model: 100 x 100 pattern of
the part below
--> 1" square x .001" part
model

It does seem to me that the "assembly method" is faster. My computer
didn't get unresponsive, but it still look a long time and was slow.

Thanks.

pope