kicad-developers team mailing list archive

["Lorenzo" <lomarcan@...>]

--- In kicad-devel@xxxxxxxxxxxxxxx, "vladimir_uryvaev" <vovanius@...> wrote:

> In this case internal units make board geometry imprecise. For example 0.2 mm (represented as 7.9 mil) track with 0.2 mm clearance (21.7 mil total)does not fit in 0.6 mm gap (represented as 21.6 mil). Problem becomes global as most electronic component manufacturers using metric units as controlling units.

Actually it's not so... european and japanese are metric, american suppliers often use mils (thous) as governing unit. There also exists an unified grid which rounds up everything to 0.05mm. Here in Italy, for example, many facilities wants mixed units: it's not uncommon to have gerbers in inches, and pick'n'place metric... our flying probes importer want taxatively imperial gencad files!

> Unfortunately PCB code very deeply bound with 1/10 mil quantum.
> 
> I'm going to make some code changes to make quantum change easy. But i need community voice on this problem.
> Which new length quantum should we use?
> Which file format changes required?

There always where a discussion on the list about one month ago... look in the archives. In summary:

You have to rewrite mostly everything AND supply proper backward compatibility to old files :D a similar unit was also proposed IIRC

IMHO the best thing for a project like this is a parallel branch, while redesigning file formats (there were/are talks about that, even).

Just remember that the discrepancy isn't too serious, anyway: even most fine processes is 4 mils at most and mechanical steps (routing and drilling) have often a 0,2 mm tolerance... so until you hit laser evaporated microviasI think you wouldn't have any serious problem :D

BTW for what you need such precision? it shouldn't be an exactly cheap process with those tolerances...