I pushed an updated version of my board and now it looks like this....?

[gmdouglas]

If I push my board to a new Fusion design, everything works as expected, but when I update a PCB ( Open a new version from a colleague,  link it to design, and push -- or open the same version and update) this happens:

What happened?

What am I doing wrong?

It would also be REALLY great if I could name new designs that I push to Fusion -- instead of forcing it to be just the name from Eagle, as Eagle really doesn't have any concept of version/revision tracking.  This is a HUGE deficiency in the whole Eagle + Fusion mashup.

[jorge_garcia]

Hi @gmdouglas,

I hope you're doing well. Could you provide more information on what you were doing when this happened?

Did you add any new components to the design before this happened? Any info you can provide about what your were doing will be really helpful.

Best Regards,

[gmdouglas]

Hi Jorge,

Here's our workflow:  I'm working on verifying a design that a colleague is working on.  He's making all sorts of changes to the board, and sends me archive zip files of just the board and schematic files.

I open the zip files, and then open the board/schematic in Eagle, and then link it to my existing design for the board in Fusion.    I then do a push to Fusion (updating the existing design) ,hoping that when I update the board, my assembly design containing it (joints etc.) will mostly be intact,  but the board gets mangled.

My colleague has an older version of Eagle ( pre-subscription, pre Fusion ) and I'm using the latest.  I've also tried changing the board slightly and saving in Eagle 9 -- but the behavior is the same.

If I push the board to a new fusion design, everything works just fine -- but subsequent updates of that will also exhibit the behavior.

There have been a few component changes throughout, and many many changes to cosmetics, routing, board shape etc, including the corner radius on the board's outline.

This behavior has been quite consistently happening for the last few revisions.

If you need to see the files, I can send them to you privately -- and if there's a better way to collaborate with another Eagle 7.x user I'm all ears.

Thanks!

[yqliu]

Hi @gmdouglas , could you please share your files to me so I can look at what is wrong? My email is : [email protected]

[gmdouglas]

Files sent. Thanks!

[yqliu]

Hi @gmdouglas ,

Thanks for sharing your files otherwise it's hard to tell what is wrong.

The problem is there are two concentric circles in package MM70-314B2-1_SMARC and SMTSO-M2.5 and both of them are on dimension layer. In this case, the software can not tell which circles are part of the the expected outline. To fix the issue, you can remove the outline circles out of MM70-314B2-1_SMARC (or SMTSO-M2.5).

Richard

[gmdouglas]

Thanks Richard for the workaround.  I'll look at those footprints.

This is a pretty serious bug though - since this doesn't explain the thickness of the board/cylinder in the model nor the absence of the other features. The board thickness should never deviate from that specified in the DRC setup.

Additionally, the concentric circles in the footprints aren't the size of the cylinder produced in fusion -- making the problem very hard to diagnose!

Should I take from this the the 'dimension' layer is wholly reserved for the board's outermost extents?  It's not forgivable that the import can't find the outermost closed features itself and turn inner ones into voids/holes.

It would be great -- and maybe a little more useful if the push to Fusion treated the design just like a PCB manufacturer does.  It seems (based on some other problem I have with holes not being holes/round ) that maybe the conversion process is more complex than even the physical process is.

While the 3d features are already quite an asset, it would be even better if Fusion could use virtual drills/machine tools following tool paths on virtual material and present a manufacturability assessment -- instead of trying to create a much less simple model however it does now.  Fusion's quite good with understanding the constraints/process of sheet metal and machining -- and I'd love to see it understand the PCB process a lot better --since it's mostly a 2D combination of both ideas. 

Additionally, the 3D asset becomes less valuable as trust in it erodes due to these sorts of errors.

Please let me know when this bug might be fixed!

[yqliu]

Hi @

Hi @gmdouglas , thanks for the feedback.

Can you elaborate the issues of the board thickness, size of the cylinder vs the circles, and the the absence of the other features?

Regarding to the dimension layer, yes, you can think the this  layer is reserved for the board outline. You can use circles on this layer to cut holes from the board but it doesn't make sense to put a hole inside the other hole, right?

Did you try the CAM workspace? It works with 3D PCB too.

[gmdouglas]

Hi Richard,

Take a look at the picture I posted in at the top of this thread.  There's a circular body about 3cm thick and ~80cm in diameter between the 2 planes that have the board's top and bottom texture.  How Fusion made it this way because of a redundant hole in the board is puzzling.  The rest of the holes, and the square outline are nowhere to be found.  This is the thickness/feature issue!

As for holes in holes, it's actually a very common solution -- and NO hole in the main PCB should cause the entire workflow to explode.  They're holes (While circular  -- they're not just circles as you describe them!) and should be treated as though they're drilled with a drill bit (like the hole tool in does in Fusion)!  The Eagle PCB file has tool size and coordinates embedded within  -- and it should be properly used by Fusion as it generates the body. It's clearly not doing that now!

I've used redundant holes to enlarge hole features without polluting my library with extra special case footprints.  Additionally, I've also overlapped footprints to accommodate different connectors or through-hole parts -- resulting in many redundant hole situations.  If I decide that I want 2 drill hits there, for whatever reason, it shouldn't make my 3d model unusable. This is a perfectly normal solution in PCB design for a number of cases -- and should never blow things up as badly as you see in my post.  This may also fix my other issue about _some_ holes not being 'hole' features after import -- and impossible to use for joints.

My previous suggestion about the Fusion side looking at it as  a PCB -- and understanding how PCB's are made -- with their own design criteria -- instead of just a fancy 3d body -- would be advantageous -- and would avoid time eating problems such as this one!

Our design may have features that are unusual -- but that doesn't mean it's wrong given the constraints of PCB design and manufacturing. 

I insist that this is a huge bug that needs serious attention!

[yqliu]

Hi @gmdouglas ,

I can see the board was not created correctly but I'm not able to reproduce the thick circular body issue. I will let the team try in different environment. I agree it's bad that a hole made the whole 3d PCB not usable and we will definitely improve it.