ID: | 4404 |
From: | ge...@igor2.repo.hu |
Date: | Tue, 29 Sep 2020 17:24:21 +0200 (CEST) |
Subject: | Re: [pcb-rnd] Back annotation |
in-reply-to: | 4403 from Bdale Garbee <bd...@gag.com> |
On Tue, 29 Sep 2020, Bdale Garbee wrote: >Evan Foss <evanfoss@gmail.com> writes: > >> I think the output in the schematic capture program should look more >> like the drc window in pcb-rnd. > >That's an interesting thought. > >In the absence of working back annotation, Btw, we do have working back annotation, see the demo video from 2015: https://archive.org/details/pcb-rnd-back-annotation It doesn't need a modified version of gschem since their last release. The feature is in standard pcb-rnd since 2015. In gschem it looks remotely like Evan described: a list of things the user needs to fix on the schematics. Just not in drc-like preview window but "jumping there". And mostly in textual form, because gschem doesn't have "rat lines". >it's not uncommon to see >something you want to change in the layout, go to the schematic to make >the changes,then forward annotate the netlist update and see the >changed rats. I recall using systems with some level of functional back >annotation where "simple" changes like which of the 4 gates in a 7400 >series gate package were used where could be trivially updated in the >schematic just by changing the section designation and letting the >resulting pin number changes ripple. I can imagine how the changes I >personally am most often making, which are pin choices on SOCs or >connectors, could back annotate fairly cleanly. ... as demonstrated by the video above! The big time saver in this setup is not that anything'd happen automatically on schematics side. Gschem is way too weak for that. And no, lepton didn't make any progress in fixing the old design errors on this either. In fact lepton is a step-back on the back annotation department, as it has even less support for it than gschem, while having the same fundamental limitations in the data model (e.g. not knowing much about nets). The big time saver is that pcb-rnd does understand that a back annotation is happening, so once you fixed up the schematics and do the next forward annotation (with or without other changes), pcb-rnd will pick out the chanegs from the forward annotation that are resolving back annotation requests and remove them from the board's "local netlist modifications for back annotation" pack. From another point of view, the time saving is that you don't need to interrupt your pcb layout work and/or think aside/ahead: when you see swapping pins or replacing footprints helps, you can do it immediately, without having to worry or remember - pcb-rnd does that for you. So it's a bit like the manual method, in the sense you still need to go and fix the schematics by hand. But it's very much automated in the sense that what needs to be fixed is transferred automatically and whether it got fixed or not is evaluated automatically. If you don't fix something in your forward annotation, it will be exported again with the next back annotation. Which means the error prone, boring data transfer between the tools don't happen on paper, but the part that needs intelligence/creativity (how exactly you want to execute the changes on the schematics) is not done automatically. Regards, Igor2
Reply subtree:
4404 Re: [pcb-rnd] Back annotation from ge...@igor2.repo.hu