ID: | 3708 |
From: | ge...@igor2.repo.hu |
Date: | Mon, 27 Jan 2020 03:51:23 +0100 (CET) |
Subject: | Re: [pcb-rnd] reversed missing layer export warning |
in-reply-to: | 3707 from pc...@cuvoodoo.info |
replies: | 3734 from pc...@cuvoodoo.info |
Hi cuvoodoo, (sorry about the late reply, I was real busy with a huge code cleanup, will post about that soon) On Fri, 24 Jan 2020, pcb-rnd@cuvoodoo.info wrote: >On Thu, Jan 23, 2020 at 06:10:28AM +0100, gedau@igor2.repo.hu wrote: >> As of r29239, I've fixed and documented both of these, please test and >> ACK! > >ACK thanks! <snip> >I tested okempty, okempty-group, and okempty-content. >they work like described. >thanks a lot for the upgrade. > >should the CAM rules be updated accordingly? >I would mark *-copper as okempty-group, I agree with this one; are you willing to contribute this as a series of svn commits? >and *-silk/paste as okempty-content (or even just okempty). >board without silk (replaced with marking on copper) and paste (THT only) are not unusual. >no mask is unusual though. I disagree with these. We shouldn't use okempty-content (or okempty) in default CAM jobs. Rationale: our layer system is flexible. If you know your board is not going to have top silk, it's best to delete the top silk layer group. If you ever place any thru-hole or smd part, your mask will be non-empty. If you know you are going to fab the board without solder mask put on it (rare), you should rather remove the layer group - it's better than sending the fab a valid mask gerber and then explain on a side channel that "please ignore that and don't put any mask on the board". If you ever use smt, it's very unlikely that you would have empty paste layer intentionally. If that's the case, that's the uncommon case; if you are using the cam jobs designed for the common case, you _should_ get a warning and decide if you just made a mistake or you have the rare case. If you have the rare case, you have multiple options: 1. delete the paste layers, making it explicit that your assembly won't be reflow, and you won't ever order a stencil 2. if for some reason you are required to pass on an empty paste gerber, e.g. your fab's process is broken, that's a special case you should handle with a new CAM job 3. or especially if it is a temporary situation, read the warning and think it over and accept it; this will help you catch a major error at the end: with using custom footprints you may forget to specify paste shapes and won't notice it for long because the paste layer is normally turned off; as a last resort you are going to figure this from the export warning for empty paste layer. More generally: our layer system is capable of representing any board stackup now, there are no special cases, no "but you have to have a top silk layer because of file format or padstack or whatever". We also have super-easy ways for creating and removing layer groups (e.g. right click context menu in the layer selector widget that is arranged to represent physical setup). So there is no reason to keep a stackup in your board file that won't match your physical board's. If you won't have a physical layer on your board, just delete the corresponding group. For a single sided board (rare), just remove the top copper group. So I'd rather keep warnings to drive users to set up their layer stacks properly, instead of keeping layer groups empty they don't really want. In this setup I think I'd have more okempty-group in the default CAM jobs but I would have okempty-content (or okempty) only in some very special cases, e.g. when we see specific fabs yelling at us when sending a through-hole design (with no stencil) without paste gerbers. Regards, Igor2
Reply subtree:
3708 Re: [pcb-rnd] reversed missing layer export warning from ge...@igor2.repo.hu
3734 Re: [pcb-rnd] reversed missing layer export warning from pc...@cuvoodoo.info
3738 Re: [pcb-rnd] reversed missing layer export warning from ge...@igor2.repo.hu