Hi,
I have found a bug which appears in pcbnew in which the imported netlist shows the proper connectivity using “show netlist” but displays incorrect connectivity, both missing rubber bands and rubber bands being present where there is no connectivity. The following images document the bug as it appears in pcbnew.
Are there any workarounds?
Bart
See my post in other thread - this looks more like a render artifact, than any data base flaw.
Did you check other canvas views ?
Hi PCB wiz
I had found that page, but when I click on the “report a bug” link it takes me to a Ubuntu site - report it here anyway?
Thanks for the help!
Bart
Hmm, when I click report a bug, I get
https://bugs.launchpad.net/kicad/+filebug
before reporting, you should run tests to see if it is a render issue, as the net highlight test indicates the Data base is intact.
Then, see if the render issue is the same on all canvas.
If it is only on legacy, it may not be worth reporting as legacy is being retired in the long term.
You need an account with the bugtracker/repo there and afaik that system is run by canonical I think… thus the ubuntu connection.
As for your bug, it’s just a render error, it will not let you connect different nets. Also as it’s the legacy canvas they won’t be doing anything about it.
I’m not sure how you would expect that to be rendered anyway, it would always appear that the 4 ‘V’ pins are all connected together. And there is no missing connections, your caps are connected with yet another overlapping connection at the ‘V’ pins. KiCad does not always draw the ratsnest as you would expect at the best of times and it will change once you start routing. It is a guide and nothing more. Go ahead and start routing the board.
I routed the board successfully using openGL, and I just read the thread about the missing features under openGL that the legacy renderer has.
I missed the old “drag end” tool in my old pcb software, and found the “manual” routing not manual enough for my taste. I found my self trying to to figure out how to “outsmart” the software. I like to route the track not only to the pad, but also around the pad so that through-holes don’t narrow the copper on a long trace every time it hits another pad. If anybody has a suggestion on that (something I haven’t stumbled on yet), I’d love a pointer\link or a quick suggestion.
Thank you all for this cool piece of software.
Bart
Did you try the Shove Router ?
To me, that is the best method, and it seems predictable and smart enough.
Not sure I follow.
There is a net-class width you can define, and select when routing rats lines.
That allows some by-net control of the starting width.
Besides that, you can route like-width-tracks in order, to avoid too much ‘width ping-pong’.
I’d avoid running traces right across a pad, as a lazy way to avoid width-define, as the database checks for trace-ends when doing connectivity checks.
I’m not sure what you mean by this or when you say you try to “outsmart” the software. If you are referring to how it snaps to the center of a pad, in OpenGL canvas you can hold down the “shift” key to prevent it from snapping to the pad. If that doesn’t help maybe you could post an example to explain what you mean.
I have tried using the shift key, and yes that helps some.
Thanks!
Bart
