Creating Components - How is this considered rational work flow?

[jwilkins8475]

It seems a lot of the video tutorials are out of date. Traditional documentation seems hard to come by and completely unofficial, created by unaffiliated third parties. Still, it's not impossible to figure out.

1. Create new library.

Well, I don't really want a whole library. I just want to make one component. But I have to save it to something, so I guess that makes sense.

2. Create a symbol.

OK. This part is all I really want anyway, and it's kind of fun. Easy enough.

3. Create a footprint.

I don't really want this, but it seems sensible to implement.

4. Create a package.

No, no! Wait! Why is an entirely different environment opening in a whole new tab?

Oh, I see, I can import a 3d model if I have one. That's easy. Oh, but I have to save it before I can insert the model. That's odd. Why save an empty file? Whatever. OK. Saved. Insert model. OK, yep, there it is. Save. Oh, there's already a footprint with that name. Right. I want this model named the same as the footprint named the same as the symbol. Because it makes sense that a model named MODEL would be part of the part named MODEL. But I can't do that? OK. I'll change it to "MODEL-2". Save. That worked? OK. Done.

That was easy, I guess.

Soooo... let me go back to my schematic and open my brand spanking new library and there's my component aaaand why doesn't it have a symbol or a footprint or a model?

Well, maybe if I go and double check in the handy and very easy to use Library Manager. Yep. There's my library. OK, there's my component. I can open it. There's the symbol I had fun making. There's the footprint. There's the.. oh, wait. Where's the 3d model? OK, it's right there in the folder I made. It has the name with the "-2" and everything.

Why isn't any of this working?

Why does something I just created appear in the library manager as 2/3 finished component?

Why does the 3d model not attach to the symbol/footprint?

Why does the component have none of any of the above when I try to place it in my schematic?

Are there any up-to-date documents - preferably in the format of numbered (or lettered, or even just realistically sequenced) step-by-step instructions on how to do this?

Why is this so convoluted? Surely, there's a better way.

[jorge_garcia]

Hello @jwilkins8475,

I would recommend going to through this series of tutorials they are number 1-5.
https://www.youtube.com/watch?v=xNIEXCimRSg&list=PLmA_xUT-8UlKE-U-eEqrkNEI7rd1fUnLY&index=1

It will clear up most if not all of your issues. From your description, you are missing the component which is what links the symbol and the 2D footprint. You will always need to define a component even if you only want the symbol. The other point is that everything in the library must be connected explicitly, having two objects with the same name doesn't automatically associate them.

These are the 2 main things that have gone wrong in your current work.

Let me know if there's anything else I can do for you.

Best Regards,

[jwilkins8475]

Hi Jorge,

Thanks for the reply. Those are the videos I watched to get me where I'm at. I'd rather just have a written guide. I was born long before 2000, and I learn better by reading than by watching almost an hour's worth of videos.

[cory_]

Just to offer some commiseration, it's certainly not just you. The electronics side is sorely under documented and what little documentation there is often lacks images, or is even in some instances, out of date.

I strongly suspect it's more an issue of the workflow of Fusion electronics releases not including documentation as part of product delivery that plays such a strong hand in the disorganization. I imagine it's a very difficult task to remedy what I assume is now significant technical debt.

I also share your feelings on the difficulty of many processes that feel like they could be so much easier. At some point I started just creating a personal list of issues I have, or work flow problems I feel are significant, and its starting to become pretty long 😐.

In my case the only component I've made was a FT2232HPQ because I found no ECAD models available for it.

I preprocessed a lot of data, used OCR to get them into files, made a python script to organize pin names with their numbers, and had it ready to go in a key value, csv, or whatever I assumed Fusion would accept only to be hit with an interface where I had to manually select every single option individually. It was a thoroughly painful experience and what made it hurt more is I could easily envision small UI changes that could have made the process a breeze. Simply having a double coloumn text entry with a list of remaining pin numbers and the ability to paste key value pairs would make this orders of magnitude more usable, and I feel its the sort of interface addition that could me made with not too much developer time. I had half the mind to see if the python API would let me do just that (As I've automated some things to do with CAD before with fusion, like auto renaming bodies), but at that point I was so tired and didn't have much hope with the api being as flushed out as it is for cad, so I toughed through it.

Ultimately, I must agree that Fusion is simply not built to be the originating place for components, because I can't imagine this method of data entry is a good use of man hours, yet no faster scripted or CLI method is documented as far as I could find, so I can only concur that it is not very user friendly/is not the right choice if you will be needing to create your own components frequently.

[jwilkins8475]

@cory_I don't often have to create components - this is, I think, the second time in 3 years. Both times led to lots of lost hours. The first time, I did accidentally end up with the footprint I needed (but nothing else), so I stopped there and chalked it up as a win.

I agree with the rest. I'm sure they made the choices they did because it's what works under the hood, but that doesn't change the fact that doing something like this should flow as easily as some of the more mundane parts of the design process.

[jorge_garcia]

Hi @cory_ ,

Dang dude, you really went all in on that. I agree and advocate for getting Electronics into Fusion's Python API. With said, once you had the CSV you could have used smart paste to get that info in Fusion.

https://www.youtube.com/watch?v=0wNLXXT_jsA

With that said it's not necessary to go that deep down the rabbit hole. Often you can re-use existing footprints and only worry about making the symbol which is much simpler. We keep a list of user requested improvements for whenever we can prioritize the library authoring experience.

Thank you both for your feedback, I'll pass it up the chain.

Let me know if there's anything else I can do for you.

Best Regards,

[cory_]

Hi Jorge, and thanks for the effort. Unfortunately, I have already used that pasting tip for the initial placement of pins. It actually initially gave me hope that I could do the same for the next part of the process which is the section I was referring to. What I was referring to as laborious is specifically the connection part where a footprint is attached to a symbol.

As far as I was able to find, the only way to do this involves a UI which, while serviceable with smaller parts, becomes tedious as the number of pins increases, especially with multiple repeat pins. A table allowing for key-value pairs to be pasted, or some API to avoid manual entry, could greatly speed up this workflow.

In fact, in exporting the footprints and symbols, I discovered that the exported format in XML seems relatively ready for data manipulation. If there were some sort of API access, one could create a plugin or option that took in the suggested key-value pairs, aligning pad values with the same pin names and collating them into the same connections. I imagine if I were more invested, I could write a Python script to handle this using said XML formatting, though it would be less reliable than an officially supported API, given that the file format might change without notice.

Regardless, I do think there are a few options for improving this workflow, as mentioned in the previous comment.

It's also good to know that the Python side isn't there, as I can imagine if I had decided to figure out a way to automate it, the absence of official support would be difficult to decipher conclusively through documentation.

[jorge_garcia]

Hi @cory_ ,

Thank for clarifying. Those connections are scriptable through ULP, which is not as nice as something more standard like Python. There is a ULP that can actually do some of what you are describing. It's UI is pretty horrible but it does work. Go into the ULPs and type bsdl, only one result will come up it's starts with the word make. Check it out, it can make a completed part from a BSDL file which comes in handy for things like BGA or other high pin count components.

It would be nice to have something built-in and native, I'll pass your comments along.

Best Regards,