Thanks for both of your comments!
This footprint appears to have both an SMD pad (to get the paste) and a through-hole pad (to get the hole) stacked on top of each other. But the paste in the SMD pad appears to be solid (there is no hole in it), so if I’m supposed to avoid getting paste in the holes, how does this avoid doing that?
Strangely, it seems to remove the
.pdf extension when I follow the link, and it gives a 500 error. But if I manually add the
.pdf extension in the URL bar, it seems to work.
The part I am looking at is TLC5947. (It’s an HTSSOP-32, which does not exist in the library, so I’ll have to create my own footprint anyway, but I was looking at the HTSSOP-28 footprint for guidance.)
On page 25, it shows their recommendation. It shows a 8x4 array of vias, and the paste is placed in a squarish area about 4.36x4.11 mm. (Although it says the exact area should depend on the stencil depth, which I guess I need to find out from my assembly house, which is probably going to be MacroFab.)
The datasheet doesn’t say anything about breaking up this square into smaller areas or avoiding the vias. The SLMA002 document linked from the datasheet does mention (on page 9) capping the vias with soldermask to prevent the paste from flowing in. How would I do that, exactly? It seems like I would need to make a circle of soldermask, and since the
F.Mask layer says where to not put soldermask, it’s easy to create a circle of not-soldermask, but the opposite seems harder.
After I create my footprint, should I contribute it back to the KiCad library as an HTSSOP-32 footprint, or should I specifically call it a footprint for the TLC5947, since other HTSSOP-32 chips might have different requirements?