How about some fuzzy DRC: 

The DRC checks for silk within either a maskclearance/2 in either the x or y
direction.  It does this by computing the refdes text length and creating a
box representing it's extents.  Then you add a fudge of 2 mils to see if it's
getting into the mask apertures.  Thinking about it, it's not so fuzzy.

Then a math-heavy correction or check could be done later.  Oh, wait, maybe
what I recommend is 'math heavy'?

Phil

DJ Delorie <[EMAIL PROTECTED]> wrote:

> 
> > I don't know too much about the gerber file format... but I have
> > access to a commercial EDA program that cuts the silk layer using
> > the solder mask. This seems what you have coded, but I never had
> > problems with it.
> 
> Well, there's two ways to cut silk.
> 
> 1. Use a negative polarity layer to undraw it.
> 
> 2. Modify the silk itself to just not draw there.
> 
> The second is a better solution, although you have to calculate the
> intersection of all silk lines with all pins/pads and do the heavy
> math to figure out how to break up the silk into segments to you just
> happen to not draw over the pins/pads.
> 
> Negative polarity undraws was much easier to implement, although it
> seems the easy way out isn't the right way after all.
> 
> Maybe we could do the heavy math on user command, and have DRC
> complain about it.  Combining that with actually drawing the silk over
> the pads on the screen so the user sees the problem might be
> sufficient.
> 



Reply via email to