open-source-silicon.dev

image.png

does anyone know why the sky130 hd tap cells have a seemingly-unused psdm region next to the tapped nsdm region in the n-well, and equivalently an unused nsdm next to the psdm in the pwell?  (here is n well)

nsdm.png

continuity. standard cells are often placed in rows, with the alternating rows being flipped on the X-axis. The seemingly unused nsdm/psdm regions overlap. It is possible that 2 tap cells might almost align vertically. If the seemingly unused nsdm/psdm regions were missing, there could be spacing errors between the nsdm/psdm in the normal standard cells.

but you can achieve the continuity if you just extend the tapped nsdm/psdm region, you don't need the alternate to be there?

oh i've been staring at a rotated cell for a long time

i think this is confusing because i don't understand the role of the `psdm.drawing` and `nsdm.drawing` layers in this scenario. the `nwell.drawing` is there, so there is continuity to neighbouring cells which draw `diff.drawing`, but instead sky130 cells have a separate `tap.drawing` with the opposite `*sdm.drawing`

i guess it does make the tap explicitly n+. ok