kicad-developers team mailing list archive

[jean-pierre charras <jp.charras@xxxxxxxxxx>]

Le 07/12/2010 01:02, Vesa Solonen a écrit :
> On Mon, 6 Dec 2010, jean-pierre charras wrote:
>
> > Pay a particular attention to a new filling zone algorithm.
> > I need volunteers to test this algorithm and know if it is better than the current one (i.e. if no issue found).
>
> New fill leaves thermal stubs between SO-package pins, that old fill did not. In the old stub-removal code there was some fudge
> factor for corner cases if I remember correctly. So maybe the new version is just more accurate. I havent looked at the code, but
> the stub issue is annoying and may need too much work to solve properly for this release. Logic to prevent stubs is something like
> "no thermals to islands smaller than minimum width squared" or so. This issue doesn't seem to violate design specification, but
> definitely worsens manufacturability.
>
> -Vesa
I am not sure this is due to the new filling algo, but there are some (minor) changes in stub code (for both algo).
Thermal stub code was recently modified and it is now common to the "old" and "new" code for filling zones, and has slight changes 
(with some very minors bugs and a serious bug fixed).

Can you compare with a very recent build.
(Warning, use of new algo is now the default when creating makefiles (option DUSE_BOOST_POLYGON_LIBRARY=OFF must be added)
Or can send me your board ?

My concerns about new code is more erroneous filled areas (forgotten areas and/or bad shapes).

--
Jean-Pierre CHARRAS