Anyone noticed, or has an explanation for these artifacts (seems like one artifact, occurring on several instances) in the Gerber files. Look at this screen shot:
This is the gEDA Gerber viewer — however, KiCAD’s Gerber viewer also shows them (incidentally, for this particular board I can’t reproduce the artifact in KiCAD’s viewer).
For the left-most marker; it looks like the trace that comes perpendicular to the pad is thicker than the pad, so that it goes past the right edge of the pad a little bit. However, I confirmed (double- and triple-checked on pcbnew) they’re both 0.2mm. Moreover, you can see the three right-most markers; those three pads have nothing connected to them, and it still shows the little bumps at the right of each pad.
For the circled part, that looks like the trace that connects to that pad is not aligned — however, I checked on pcbnew, and it is — Start point X: 113.4; End point X: 113.4; pad properties: Position X: 113.4
When zooming in, the artifacts disappear — in any case, they show at some zoom levels, and it seems to depend on the actual position; I just went to this view, and had to zoom in six steps (mouse-wheel steps) for the bumps to disappear; then, zoom back out with the mouse wheel also, and now just after two zoom-in steps the bumps disappear.
Notice: this behaviour/artifact by itself should not cause any trouble in practice (even if those bumps were actually there, they’re so microscopic that they fall far below the “noise level” of the PCB manufacturing machines). However: (1) it could be a sign of a more subtle bug that could manifest and cause real trouble under certain additional conditions; and (2) it’s still a “paper cut” in the sense that it is visually distracting and stressful.
The footprint is taken from KiCAD’s built-in library and modified slightly. I’m also posting that file, just in case (not sure whether it is already embedded in the pcb file):
Let me know if you need any additional details or files.
[[ EDIT: forgot to comment: though the evidence, at first glance, may point to “this seems like a bug in gEDA’s Gerber viewer”, I’m quite sure that I saw this yesterday when playing with KiCAD’s Gerber viewer; it may have been on my home PC (I’ll try to confirm tonight when I’m back on that PC); both PCs have the exact same version of KiCAD, installed from the something-reynaud repository (i.e., not compiled from source); a difference is: this PC runs Ubuntu 18, whereas my PC at home has Ubuntu 14 ]]
D11 is the spot circle of radius = pad width. D0 is the surrounding polygon. gEDA is showing you the spot larger than it is (0.2mm) – Or, more likely they are not drawing the polygon correctly
I’ll try to double-check when I get back home — it could be my memory playing tricks on me, but I was sure I saw this same artifacts (not for this same file) on KiCAD’s viewer.
It is also curious that I never noticed any glitch when viewing Gerber files produced by gEDA pcb — I guess that can be easily explained, since there may be multiple ways to represent the same layout in Gerber language, so maybe gEDA pcb generate Gerber files in a way that shows perfectly well through gEDA’s viewer.
Nope, can’t find the problem with KiCAD’s viewer on my home PC. In the PCB I’m looking at, there are things like traces that go through a slightly wider pad; maybe I saw one of those and it looked familiar so I thought it was an instance of that same issue… or, it could have been my imagination! :-\
Oh man, I love this forum!!! I’m here learning new stuff about gEDA!!! (I had no idea that gEDA’s Gerber viewer had several rendering levels!!! )
Anyway, at High Quality, the little bumps no longer show (in the screenshot I posted, the four arrows) — however, the misaligned trace still shows (the one that I circled in a dashed line). And same thing: the artifact only shows at some zoom levels (higher zoom-in levels tend to make the artifacts go away).
I’ve been starting to use KiCAD’s viewer — however, functionality-wise (i.e., buggy behaviour aside), I find myself finding some (most?) of gEDA’s viewer aspects being clearly superior (which I guess we’re lucky that Gerber is rather outside of the workflow, so it’s not a problem to use independent tools for it).
In particular, what I’m disliking in KiCAD’s viewer is — and please correct me if I’m getting these wrong:
Very limited stacking of layers — with gEDA, as you can see in the screenshot’s bottom-left corner, I can dynamically rearrange the order of layers (I select one and click on one of the arrows to move it up or down).
With KiCAD, I can only select one to move to the top — the remaining ones keep the order that they have as per the layer manager. (I guess transparency might be a saver; but only to some extent)
Opening a set of Gerber files is truly slow and painful. Not sure what exactly I would suggest or what to pin-point; but the thing is; the menu is overcrowded (with drill files being separate from Gerber, and open and open recent being separate items — with the four combinations there), and it requires several clicks to get each layer loaded.
With gEDA, I can go to the File Browser (Nautilus, in the case of Linux/Ubuntu), select several files, right-click and Open With… gEDA Gerber viewer. I can’t seem to be able to do the same thing with KiCAD; the viewer (or pcbnew, or eeschema, for that matter) does not show as a standalone application; just KiCAD itself; but that seems to be the “main” KiCAD application, which is expecting that it’s going to open a project.
To add insult to injury — after I do that, the next time I open KiCAD, I get an error, because now it kept the setting that the “last project” it worked with was the xxxxxx.grb file!!
If the Gerber files change on disk while I’m displaying them, with gEDA, I can click one button to reload the files from disk; I can’t seem to be able to do that with KiCAD’s viewer, and this relates to the second item above (having to close, re-launch and re-open all the files is a real pain!).
Thankfully, the D-codes can be hidden!!! But I’m not sure how the developers thought it was a good idea to show those by default!
I think there was something else that I’ve found not-so-good, but I can’t think of anything else right now.
Huh — how come that feature didn’t make it to 5.0.1? (I just upgraded, and am now running 5.0.1)
Regarding selection of multiple files, it didn’t occur to me to try, because I figure I wouldn’t have control over the stacking.
BTW, one advantage I see in KiCAD’s viewer w.r.t. gEDA’s viewer — with gEDA, I’ve always found a royal pain in the neck that each time I open some files, I have to individually select the colors … Call me maniac, but I simply can’t stand something that defeats the most basic common sense, which is, e.g., that holes should show in black or near-black (unless it is the only file being loaded, of course!), and silkscreen should show in white (or near-white)!! While they’re at it, why not display with a default “copper-like” color!!! (obviously, what I’m saying is: why not allow the user to specify a configuration that is persistent?!)
Anyway, at least in KiCAD, I configure the colors I want, and then since I’ve always individually load each file, I can select loading into each Graphic layer.
In nightly builds it’s also possible to reorder the layers according to their “real” order if you allow gerber x2 extensions when you generate the gerbers. Alas, not all manufacturers accept them, but you can generate one set for your own viewing.
Or you can generate a gerber job file, open it and it also seems to reorder the layers.
At least in the nightly builds this setting is saved so you have to set it only once.
BTW, the gerber job file is quite handy - when you open it, gerbview loads all gerber files belonging to that board. The downside is that it doesn’t load non-manufacturable layers if you happen to need them. But if you don’t, you only have to open the job file once for a project and as long as it’s in the Recent queue you can open all layers with just 3 mouse clicks even between sessions. As long as there is the same amount of layers, the colors should be correct, too.
I’m pretty sure it was already like that in the original 5.0.0 release; I know I’ve used it a few times, and I’m pretty sure it would have bothered me if I had to each time go and re-do the setting.
Interesting. Reordering according to the “real” order is a good thing; but I would also like to have the option to reorder arbitrarily — when viewing, one may want to focus on some details that may be easier to discern with a specific non-real layer ordering.
)