STEP file component positioning- my parts blow up!

[patmat2350]

Trying to share a complex assembly from F360 to GrabCAD via STEP export, where users may wish to rescale the entire assembly as needed.

Export to STP runs fine. Displays in GrabCAD, check. Download to local PC, fine. Upload to F360, displays perfectly. 
SCALE the entire assembly- oops. Most parts are fine, but many items, both entire components and individual bodies in the original model, fly out of position. And it seems random.
Also tried without GrabCAD- just exported the STP, and then directly uploaded locally into F360... same issue, with different random bodies. 

Is this an F360 issue? Or an instability tied to how parts were positioned in the original model? 

Again, the STEP file loads without issue, but scaling causes some random parts to fly out of position.

[jeff_strater]

can you share the Fusion design that shows this?  One possible cause is a fundamental difference in how STEP and Fusion represent transforms for sub-assemblies.  Fusion allows difference instances of the same sub-assembly to be in different positions.  STEP does not allow that, so a round-trip could possibly leave things out of position, but that should happen even without the scale.  The issue with scale may represent something else.  Does the original Fusion design scale OK?

[patmat2350]

a) Not sure how to share the design here? .f3z? Maybe the .stp would be useful? Attached.
Edit: NOT attached, because this: "The attachment's ejengelasy v177.step content type (application/octet-stream) does not match its file extension and has been removed."
Eh?

b) Can't do a single scaling of the original assembly in F360. With the uploaded STEP model, all the components can be multi-selected at once for a single scaling op. 

[jeff_strater]

yes, export the F3Z and attach it here.  I don't think the STEP file would help, to be honest, as the problem already exists by that time.

[patmat2350]

OK here is half of it, due to file size... but the problematic parts are included.

This asy is one level down, but it exhibits the same issue:  Export to STP, upload to project, all ok... multiselect all and scale all, parts go flying).

Note: The model is an unholy mess of inserted/moved/copied/jointed components, and some drawn in place relative to the main or a subasy.  But again, as the model holds its shape through the STEP translation, and only partially explodes upon scaling, I'm left scratching my head... 

[jeff_strater · Accepted solution]

Thanks for sharing the model.  I think I understand what is going on, but more importantly, I think I have a workflow that does not have this problem.  See the screencast.

First, that is a very nice model!  Nice attention to detail on that ship model.  Very impressive.

So, I don't think that the effects you are seeing have anything to do with export/import and STEP.  It seems to be because a round trip through STEP produces a direct modeling design, where your original model is a parametric design.  And DM designs behave differently from PM designs in Scale.  The reasons have to do with how parametric designs behave, and are too deep to go into here.  But, the limitation is:  in PM, you can only scale bodies, not components.  But, in a direct modeling design, you can scale components, bodies or both.

I made a guess here, based on the results I got, that when you scale the re-imported model, you are doing window select.  Is that true?  What I believe is happening is that you are getting a mixed selection - some components and some bodies, and in a non-deterministic order.  So, the scale is being applied inconsistently.  This is why, I think, when I tried this, I got different results on different scales.  

The more reliable (and much faster) workflow is to just select the root component - this scales the model from the root, and, at least as far as I can tell, correctly scales the model.

The last point of proof is that I took the native Fusion model, and converted it to direct modeling, using "stop capturing history" on the toolbar.  Once I did that, the model behaved the same as the STEP model.  Which, I think, absolves the STEP translator.

Hope this helps.  Are you planning to 3D print this model?

[patmat2350]

Many thanks... I'll take some time to digest that all, and figure out a complete workflow to help make it idiot-proof for users. For example, I'd not want to lose design history on my original model! 
Yes, I window selected the bodies and components when scaling the re-imported STEP model... selecting the root looks easier and it works!

A printed version could be in the offing, but not directly from this assembly... I would use this basic geometry to define printable parts. 

Thx again!