open-source-silicon.dev

image.png

Hello everyone, I was studying the layout of some cells from sky130 pdk and I would like to know why all of them have some nwell portions outside the yellow box? I play with the layout (fig 2), and see that without them (red boxes) there is no DRC violation. Same thing with the blue box, where in this case there is no nwell, I filled it and no DRC error. Thanks in advance.

The abutment box shows where cells should touch in an array---it is what is used by digital place and route tools to place the cells into a standard cell layout.  The actual layers in the cell may extend beyond the abutment box to satisfy various width, overlap, and spacing rules.

All right, thanks Tim. Regarding the top part where there is no nwell (yellow box), I guess filling or not with n-well was optional and the design team adopted not to fill that part. what do you think/know?

inverter.png

Your "yellow box" here is not really the abutment box;  attached is a screenshot showing the abutment box (inner white box) and extent of geometry (outer white box).  The abutment box runs through the centerlines of the power and ground rails, so when cells are abutted, the rails overlap.  So do the nwells.

What is this yellow box? Is it a boundary box? How do I use it?

Currently I'm designing inverter-based amplifiers, but I need extended nwells to pass DRC. It is not area efficient, so I must overlap subcells

<@U01QTMG2K8R> I draw that yellow box with my screenshot tool. I don't know if I understood very well what you want to do, but If you need to extend the nwells, could you use filler cells?

<@U016EM8L91B> One question, Fig 1 is my inverter focused only on verifying all DRCs, so the n-well is as large as required to satisfy the 0.18 um with respect to diffusions.
You mentioned that the actual layers may extend beyond the abutment box to satisfy the overlap. My question is, how big should I draw the n-well so that when doing p&amp;r I am sure that the tool will create continuity in the n-wells? Is Fig. 1 ok or should I fill in the empty space as in Fig. 3?
I have measured in sky130 hd std cells the n-well outside the abutment box [Fig 2], are those extent dimensions mandatory?

In principle, as long as it reaches to the abutment box edge, it will be okay;  there is no particular reason to require the nwell to satisfy the minimum rule within the cell (in Magic, where the DRC engine is interactive, the DRC error will show up as long as you are viewing the cell by itself, but will not show up in the top level view of a placed and routed design).

The idea may be that designers (especially mixed-signal designers) may use a handful of digital cells outside of a P&amp;R environment, so having the cell clean by itself is useful.  By contrast, I/O cells are often not DRC clean by themselves, because they would never be used outside of the context of a full padframe.