topic Re: Extrusion (Press pull) Bug in Fusion Design, Validate & Document Forum

Extrusion (Press pull) Bug

ericschimel — Fri, 29 Apr 2016 01:15:36 GMT

Hey all, I'm running into an extrusion bug. Check out the screencast:

https://knowledge.autodesk.com/community/screencast/c8a6f240-0589-4cae-8be7-243a25ade8de

Basically what is happening is that I'm using a rectangular array to to create a bunch of shapes. Then I'm using an extrusion to tie them all together and make them one body. This all works fine the first time, but the problem is when I change the amount of shapes in the recangular array.. Sometimes they rejoin with my connecting body, sometimes they do not, the behavior is incosistent.

Check the model here: http://a360.co/1TzlsqH

What do we think? Bug, or am I not modeling right?

Re: Extrusion (Press pull) Bug

TrippyLighting — Fri, 29 Apr 2016 14:25:21 GMT

Looking at the screncast I see that there are a nuber of irtems in the timeline highlighted in yellow. these indicate warnings, often lost refference geometry.

Try to fix these as they often cause problems later in the timeline. HWether that is really the problem in this case is hard to say, I am at work, away from my Fusion 360 station and cannot check the model at ths time.

Another tip is to occasionally run "Modify->Compute All" Sometimes the yellow the warnings don't show up immediately when they should and a "Compute all" usually exposes any wanrings.

Re: Extrusion (Press pull) Bug

Phil.E — Fri, 29 Apr 2016 17:25:55 GMT

Great advice Trippy, keep those timelines error free. You really don't want to use cached information in dynamic parametric designs.

As for this extrude behavior: This is as-designed.

The extrude/combine operation you are doing after the pattern at first only knows about the bodies it joins.
These bodies all have (for lack of better description) names and addresses.

When you add more instances in a pattern by editing, you are adding new bodies with new names and addresses.

The extrude already has a list of names and addresses of the bodies participating in it's operation.
Extrude doesn't know that your intention is to include these new objects, unless you tell it to.
Thus, an edit is required for the extrude that comes after the edited pattern to tell it "use these new bodies also".

The reason this is required are the participation rules for extrude > join, which are basically to include anything that is visible in the operation that can also be joined.

This requires you to do the extrude edit your self so you can manage what is included by managing visibility of objects.
I'm not saying you are wrong to question this, just saying there are other reasons (that you enjoy without noticing) for this behavior. It's a great question and good discussion, perhaps this can one day be improved.

I hope this helps, please let me know if more explanation is required.

Thanks,

Re: Extrusion (Press pull) Bug

TrippyLighting — Fri, 29 Apr 2016 23:51:36 GMT

The real reason the multiplication of tabs does not work as in the first attempt to this model in the other thread is that in this case the tabs are patterned first as bodies and then when the center body is extruded, the number of tabs present at that point in time are joined to that newly extruded body. When the parameter is changed later in the design, the added tabs are new, separate bodies that are not joined.

The model in the other thread created that center extrusion first and then these tabs were patterned as fatures and not as bodies and that seems to make all the difference. I posted an alternative approach to this design in that thread and I've tested most parameters. It seems to behave very stable. I think I might have been inconsistend when dimensioning the tab angle but that's easy enough to fix.

I see a couple other areas for improvement.

No grounded component. Every assembly should have at least one component that is grounded.
The base sketches for the three main components are under-constrained, which is actually very easy to change by constraining them to the origin.