Hi,
I am finishing up a design, I have posted a few times here commenting on just how poor parts of Fusion Electronics actually is, it's clunky, it crashes *a lot*, it's unintuitive to the point that if you don't use it super regularly, the "weird things" you have to do leave your mind and you spend hours trying to figure out how to do something that is seemingly easy.
I'm not even at that point today, I'm at the point where I have a library, that contains a schematic symbol and footprint for an esp32-wroom-32e module, that workflow works ok. What I'm trying to do now, is go through the footprints on my design and add 3D models, sound simple?
Yes, I thought so too, but hours spent trying has to lead me nowhere, in fact, it lead me down a road where I managed to get a 3D model into a symbol which I was then unable to remove, Fusion would not allow me to undo this action, I was stuck with the 3D model (that didn't work) and I could not find a way to delete the 3D model, fusion would just tell me it's in use and that was that.
So, I ended up re-creating the symbol & footprint (well, cut and paste) and then assigned that new symbol to the schematic at which point the "lock" on the symbol was released because the previous symbol was now not in use, I was then able to delete the old footprint and the 3D model from my library which I'm trying to keep clean and organised, but Fusion has other ideas about that.
Once I did that, I then went back into my library and renamed the symbols back to the original naming and I was back to the point before I attempted to get a 3D model to work.
But now I'm back and I have no idea why it didn't work. The symbol showed the 3D model in the library browser, and when selecting the symbol in the schematic editor both the footprint and the 3D model were displayed, but pushing the 2D to the 3D board representation just throws up a warning telling me that the 3D model might not exist, I have no idea why, both in the schematic and PCB side they both show the 3D model, and in the browser, I can see my module, but alas, there are no bodies in it, it's empty.
I'm stuck and I'm incredibly frustrated why this is seemingly so difficult or unintuitive, when you go through the process you get random "save" windows appear with no explanation of what or where you should be saving stuff when you add a 3D model, there's not a hint, this is a more general bugbear with Fusion 360 in that the error messages are, for the most part, just junk because in my experience so far, they offer no information.
I must be missing something, it can't be this hard to get a 3D model to work, especially when it's showing in the library and the symbols+footprints, I don't understand why it wouldn't show. This isn't just an isolated incident, it happens no matter what I try, and I'm loathed to try other experiments because of the hassle I had removing a 3D model from a library.
When I created the package, I could see the footprint and I aligned the symbol to the footprint, everything as far as I'm concerned looks right.
The fun doesn't stop there either, the board I am working on I just had to add some milled slots too, and again in a "this is not intuitive" kind of way, the suggestion is to use the "dimension layer" for slots, but I continue and exporting Gerber and uploading to JLCPCB shows the correctly milled areas, but the CAM preview doesn't show slots in F360, and this also applies to the 3D view of the PCB, devoid of slots.
I've searched, and I've read various things and I'm still not sure if fusion is capable of showing these, some posts seem to indicate that instead of using a line on the dimension layer, I need to use a closed polygon, but I've tried that in the centre of the board with a simple square on the dimension layer and neither the CAM output nor the 3D view show any changes, all I see is the plane running away, there is no cutout.
This leads me to the next issue, the CAM export. Because I have these slots, I have to enable the dimension layer in the CAM generator, I can save this preset and it works fine if I go through the CAM processor and select that profile every time, but there appears to be no obvious way of making the "generate cam output" button use my new profile as standard, it just uses a supplied one, maybe I've missed an obvious setting somewhere, but I cannot find anything that looks vaguely like it will do this.
And then we hit the jackpot, something I've seen mentioned here quite a few times, converting bitmap to symbol.
I'm staggered that it's been years of people asking for something better than the, and I'm being kind here, abysmal method of doing it via a ULP. The settings inside that window are just terrible, I don't know why there has been a need to make it so complicated because I've actually implemented this myself in another PCB design software that I was a developer on, you only need one or another of either the target width or height of the bitmap you want, there's no need to select units, scale factors or anything else that makes it much harder to understand what you're going to get.
That window needs 2 input boxes, one for width and one for height, they're mutually exclusive so if you set the target width, the height is calculated by the aspect ratio, and vice versa. The input boxes in it are awful too, I find myself having to put in a random number so that I can then use the cursor keys to put the right digits in the right place, the input boxes won't allow you to go to an empty state, so instead, they make it far more convoluted than it needs to me.
The next issue with bitmaps is the choice of input, indexed Windows BMP? Come on guys, it's 2021, nearly 2022, there are a million libraries out there for dealing with image formats, and plenty of tools or methods to produce non-dithered images which could just as easily be used because you can reduce the actual number of use colours in the image, I use this method myself.
This pain is even worse on macOS because it, quite rightly, doesn't support an image format that is dead, so I now create my image in a vector tool, export that as a PNG at high-resolution (this is the important step because the import into fusion is poor via the ULP), so I generate the high res PNG, I then have to drop to the terminal and use ImageMagick's convert tool to convert my high res, antialiased true colour image into a true colour image, but one that contains only 2 different colours, RGB(0xFF, 0xFF, 0xFF) and RGB(0x00, 0x00, 0x00). This, in turn, goes through convert again, to take that image and turn it into an indexed BMP file, by this point we know we only have 2 colours because my workflow from vector->BMP handles that.
At this point, I can open this file via the ULP and it correctly shows just 2 colours (black and white) and now I end up scaling this image down, but to make life easier, I use decade multiples for the original input size so that the scale factor is simple to calculate.
It's at this point that you realised that it's just silly that you need to enter scale factors or multipliers and that the ULP itself should calculate these, the only thing we as users should need to enter is either the desired width or height of the resulting symbol in F360, the only thing the user needs to be concerned with is using a source image that is of high enough resolution to create a pleasing final result.
I've attached a screengrab showing the warnings I see regarding the 3D model.
