RFE: footprints using sketch tools

[djon_3V3]

Hi there,

Wouldn't that be nice if we could use the sketch tools to define components? We get footprints as nice drawings, with nice dimensions and things, and we have to input them using antique methods while all the tools are already there. 

I would say yes. Not only it gives us more time to concentrate on more important steps, but it also reduces risk of error and enhances our precision.

Thanks for thinking about it.

Have a nice day.

[jorge_garcia]

Hi @djon_3V3 ,

I hope you're doing well. Thank you for the excellent suggestion, this is currently being worked on. Hope to have some news on this front shortly.

Best Regards,

[djon_3V3]

Wee, that was fast. 😉

Thanks for the information. Can't wait for that to arrive.

Have a nice day.

[wwfeldman]

has it arrived?

i just did a sketch for a footprint for a step file from manufacturer.

i have a symbol, a footprint, and a 3d model

i want to combine them into a device

how do i move the F360 sketch into F360E footprint?

[TrippyLighting]

@jorge_garcia any feedback on this?

I discussed this idea at Autodesk University in 2017 with the developers... 

[matt.berggren]

Hi all —

We have discussed this quite a bit recently. It is definitely something for us to try and see how we might handle a few things which the 3D environment doesn’t support like pads, vias, polygons, etc.  Also since sketch lines are really the boundary of a face or an edge of a feature, they don’t have a width.  So they would have to be polygons of some sort & mirror the rounded end caps that are in ecad tools by default (which is the product of a draw aperture in Gerber and still paying the “plotter tax” from the very early days of Gerber, HPGL and film outputs).  Round ends require arcs which in large numbers can be expensive since they are triangulated in 3D and sketch is actually 3D-based.

There is nothing here that can’t eventually be solved but given the effort, we have been hesitant to make this a higher priority than addressing some of the performance concerns we had in the initial releases of Fusion Electronics and some experience gaps we knew we had, like panels & some weirdness there and of course any hard crash that users experienced.  

Since the next release has a LOT of fixes and improvements (500+) to these areas of the core, we are (re)starting this discussion.  We got as far (in the past) as even prototyping a version of this but strategic sands shifted beneath our feet and we are just getting our footing again now.

Please consider joining our insider program (early, pre-release program) and please expand on what you would like from a sketch-experience here so we can keep the discussion going.  We are eager to get the feedback.  We make decisions based on (as you’d hope) the corroboration of feedback and narrative from the users whenever possible.  This will help me sell this to the team and drive clarity which is what drives velocity.  

Thank you,

matt - Autodesk

[TrippyLighting]

@matt.berggren thank you for the very quick feedback! This is very good to hear!

I am not sure a constraint base sketch approach needs to be extended to "everything".

Below is an image of a very challenging PCB. At least it was very challenging 6 years back.

It was challenging, because no "reasonably priced" software provided what I needed. 

Creating the board outline in Eagle would have been an absolute Nightmare. So I exported the sketches I created in Fusion 360 into a DXF files. The semi hexagonal footprints for the LEDs  were non standard and very specific to this particular board, and so was the routing. 

I then used Altium Circuitmaker to create the PCBs for that project. The reason I used CM is because there I was able to turn imported DXF lines/curves into traces. For some of the PTH pads I needed the ability of having a  different sized annular ring on the top and bottom side.

The "square-wave traces on the three arms are purely decorative. Once I knew that I can convert imported dxf geometry into traces   this was easy to do.

I have not tried this lately, but I believe today I can easily sketch the and model board outline in Fusion 360, without the need to export/import DXF or any other file format. What I cannot do is to convert sketch lines/curves into traces.  

When I can design that board in Fusion 360 without obscure workarounds I think the necessary functionality has been implemented.

 

[wwfeldman]

your responses are fascinating

on a slightly different note, the package generator works reasonably well for standard packages

the computer clearly has an algorithm to build the 3d model and build a footprint, complete with 

pads. once mated to a device, the footprint can be modified. (i'm thinking changing pin 1 to square

instead of round, adding text, etc).

clearly the algorithm can set the correct layers for all pertinent things in the footprint (pads, name layer,

value layer, outline, keep outs, etc.

please provide the algorithm so users can build their own non-standard parts

[djon_3V3]

Hi.

Happy to have news on this.

Being on insider program and willing to test, give feedback, please count me in for anything you would like to throw at anyone. 

[wwfeldman]

hi matt

i re-read your posting.

"...which the 3D environment doesn’t support like pads, vias, polygons, etc. " 

i don't follow this. the 3D model generator clearly produces pads - i just used it to make

the 3D model and footprint for a Infineon (Cypress) CY8C5888AXQ 100 pin square

thin quad flat pack. it worked great.

as for vias and polygons, they get made in/by the router and the PWB design.

"...please expand on what you would like from a sketch-experience here so we can keep the discussion going."

i spent a lot of time and effort learning a little about F360 mechanical / 3D.

in all of my projects, i serve as the electronic designer and the mechanical packaging designer.

because i expect autodesk to eventually stop supporting eagle (in favor of F360E), i am using F360 for the mechanicals and electronics.

i want to draw my PWB outline in F360. for me, it can be a 2D sketch. i need the PWB outline and mounting holes to line up with the mounting posts and the box it goes in. when i design the box, i add a design for the PWB, so i know it will fit. i want to import that sketch into F360E, and finish the electronic design, then take the 3D model and back fit it into the 3D box. 

thank you

[matt.berggren]

As I mentioned, there is no hard barrier but this is not the same as saying "because I see what look to be pads after using the package editor, there are pads happening in 3D which would be something you could generate a Gerber from".  I hope that stirs things up and helps to frame this properly.

There are objects there which look like pads but they do not have a net, orientation, polygon connect style, swap information, pin map, etc...in 3D.  They dont have this because in 3D there is no downstream Editing process used that today hangs-off of that information.  The pads you see in 3D are mechanical pads - they have a name and a timeline to produce geometry.  Processes like Routing, drawing planes, generating an IPC-D-356 netlist, Gerber, ODB, etc have not historically been 3D-sourced so to add a shape is not the same as adding the ability to define the underlying data model for everything we need...So what may not be clear is the difference between the data model we need from the 2D board with the 3D PCB version of that same data.  To simply create the geometry and say "this is a pad" means that a hole has to travel with that object.  It has to cut the board.  It has to be plated.  That plating needs a thickness consistent with all other plating thicknesses from all other sources (vias, pads, slots, etc).  (If you want to know more on that, I'll be happy to share some of what we had to work out to make things fast enough for 3D interoperability...it is pretty heavy).  

Point being, all of those properties are not the same SW data model as what we need for PCB CAM that are stored in the library and defined on-creation of a pad in the 2D library and we do a lot of moving data transparently to across the editor when you create a package to ensure that 2D asset is also created but it is actually not the same as having the ability to specify all pad properties as we have in 2D.  They instead share information with one another and the goal would of course be to make that as transparent to the Platform as possible without duplicating it.  All I'm suggesting is that we are willing to look at this but the idea of sketching and extruding an object is not the same as mapping that geometry to the pin in the library.  There is quite a clever bit of work we do to back propagate / forward propagate the shape information and map that to a mechanical timeline with associativity.  

Like I mentioned, there is no reason we couldn't do something more / similar in this regard, however it is not the same as what you see in 3D is what cascades into a test point output.  Hope that helps clarify the distinctions.  We are tackling these problems because of the nature of the effort on Fusion but they are big questions that need careful examination against real workflows to ensure nothing unravels.

Cheers,

Matt Berggen

Director of Engineering - Autodesk

[matt.berggren]

To this end, we are working on the DXF/DWG translation for ECAD and looking for other capabilities here that we have suggested we would in the past but which the Raster/Vector 2D/3D transformation of data has always made more difficult.  This isn't to say it can't be solved.  We added splines to 2D however Gerber doesn't know anything about splines.  It is a draw aperture of D10 diameter <n> mils|mm.  This has always been the rate limiting factor for the big changes we would otherwise hope to make.  😞  ...We can do more here for sure though so I can attest to the fact that we aren't concerned about making this an incredible experience, just bound by slower, careful examination of these two worlds coming together.  

Best regards,

Matt - Autodesk

ps that board looks pretty cool...is that a rotary encoder of some sort?

[wwfeldman]

i don't think i said

""because I see what look to be pads after using the package editor, there are pads happening in 3D which would be something you could generate a Gerber from". "

if it is, i meant that the footprint has the pads and it works in the PWB layout and routing. the software clearly produces gerber files that have the pads, vias, holes, traces, silkscreens, etc. 

"There are objects there which look like pads but they do not have a net, orientation, polygon connect style, swap information, pin map, etc...in 3D."

what i need in 3D is what the F360E already provides - a reasonably decent representation of the real thing that i can put into a reasonably decent representation of the box, as generated in F360M.

what i want is to easily bring my 2D shape developed in F360M into F360E as the PWB outline.

[matt.berggren]

Ok, I dont think there is any disagreement here.  Just expressing that in the original post the question was about creating a part from a 3D model (if I read that properly).  This is actually a good idea we want to do.  We just need to consider "how?" to do this in a way that lets the user define the pads where the leads bond to the board.  Those will be either thru-hole or SMD and since some complex parts which mimic the QFN with a thermal "pad" this cascades to include things like polygons, vias, etc. as well as complex solder-paste stencils (which we can do with the package editor).  I use a lot of complex RF connectors and would love to trace a boundary around the leads and say "this is a pad" which is what I understood this post was about.  Same with high-power connectors, Triodes, etc - we find a lot of cool models in 3D but making some of the mating footprints would be improved from the addition of the ability to outline a pin or project the outline of the part and generate silkscreen / pads.

Best regards,

matt - Autodesk

[djon_3V3]

Hi. 

throwing out my 2 cents here.

First, my original post was because on datasheets, we normally get nice drawings with dimensions for footprints, but when we have to draw them, it's a mess. I would have liked to be able to put pads and holes and use dimensions to position/size them, instead of playing with coordinates and sizes in non-friendly frames. That would make the job of doing footprints less cumbersome and much less prone to errors.

However, now that you talk about playing with 3D meshes, here is something on the "how" of the method i envision:

imagine i have a 3D component, it is accurate and have all pins (bga, through hole, blah...). Add a virtual plane under the component and let the user place it vertically. Then, use intersections between the plane and the 3D object to define zones that are projected to the footprint plane. Then; let the user "change" zones to either pads, holes, copper pour. All zones changed to valid types would be kept in the footprint, others would be discarded.

I believe you have all algorithms for all this.

I even believe that with some clever trick ( see what components inersect also below the footprint plane) you could automatically define through-hole and enable cut-holes. Using material (if present)  you would easily be able to discern between pins/pads/casing and propose correct options by default.

I would happily test that for you, anytime. 🙂

[matt.berggren]

Thanks for that feedback!  It actually validates what I have been thinking along with the team for a while now.  My gut tells me that trying to duplicate all of the library capability in the 3D environment, though perhaps nice, would be a bit much and probably not make things faster (I have a history with RF circuits and the connectors can be a bit 'unique' depending on the application and the number of downstream units we were intended to build).  That said, I like this "convert this geometry to a..." model as it serves to address the complexity and where we can't map that directly, forces us to look at our data model and decide what to do with it.  I will bring this back to the team and meditate on this a bit to see what we might be able to do but before we do anything specific, we might just take you up on that offer!  🙂

Best regards,

matt - Autodesk

[TrippyLighting]

---

@matt.berggren wrote:

... Just expressing that in the original post the question was about creating a part from a 3D model (if I read that properly).  

---

The original request was :

"Wouldn't that be nice if we could use the sketch tools to define components? We get footprints as nice drawings, with nice dimensions and things, and we have to input them using antique methods while all the tools are already there."

This did not mention any form of 3D model, but the sketch tools.

I think that would cover 90+ of use cases where it is difficult to create  footprints, pads, copper pours etc. 

@djon_3V3 when discussing 3D geometry CAD software that usually excludes meshes. CAD software is BRep /NURBs based.

[djon_3V3]

---

@djon_3V3 when discussing 3D geometry CAD software that usually excludes meshes. CAD software is BRep /NURBs based.

---

Well, if all you noted was a misuse of "mesh" over "3d object", i guess that the remaining parts of the argument make sense. That's nice.

I would love to be able to use sketch tools to define footprints, but i would totally crave any method that would use the 3D objects i would have to model anyways to also help in defining footprints in a much faster and direct fashion.

I would pleasantly evolve from original position to the 3D method if it goes that way. And i for sure would not regret not being able to use sketching tools. 

[TrippyLighting]

---

@djon_3V3 wrote:

---

And i for sure would not regret not being able to use sketching tools. 

---

How would you model parametric 3D geometry in Fusion 360 without using sketch tools?

Usually you need at least one basic sketch.

[djon_3V3]

---

@TrippyLighting wrote:

---

@djon_3V3 wrote:

---

And i for sure would not regret not being able to use sketching tools. 

---

How would you model parametric 3D geometry in Fusion 360 without using sketch tools?

Usually you need at least one basic sketch.

---

Yeah, sure. One will always need sketching... to sketch objects. Here, the subject is footprints.

On the footprint aspect of things, if there were a method (such as i described) that would enable me to make my footprint representations off the components, i would be more than happy not to have to sketch them at all. Hope that makes it clearer.

Making 3D components is already making sure we have everything in place. As such, with some little help, it is a valid source of information to infer footprints and help user not have to do the job twice (and make mistakes).

I see having helpers from 3D components as a good way to promote accurate and complete libraries. Otherwise people will just do the footprints and ditch 3D representation (already the case imho), which lowers quality of libraries compared to competing alternatives. Not only that would be good for autodesk but also for users. Everybody wins.

Also, it would be pretty awesome in demos to drag a 3D component and see footprints and cutouts being automatically proposed, immediately. Talk about some wow effect.