@Tim Edwards The DRC rule in magic for gf180mcuC: Diffusion contact width < 0.23um (CO.1 + 2 * CO.6) - I don't understand why we need 2 times CO.6 there, could you please explain?

@Philipp Gühring I’m sure Tim will give a much better explanation, but cuts (contacts/vias) in magic are not actually shapes that are written to gds, but rather regions that are filled with contacts/vias when written to gds. The drawn shape is converted into metal and minimum sized cuts offset from the metal edge by the minimum overlap. Thus the drc rules in magic require the drawn shape to be at least the size of the minimum cut plus the minimum metal overlap.

But the minimum metal overlap should be on the metal layer, if we are enlarging the contact itself, the metal needs to be enlarged twice (and then any other things that need spacing to metal also need to be shoved further away)

In magic, the contact layer exists on both metal planes.

Hmm, so in gds2 the contact would need just the CO.1 size and in .mag it needs to be CO.1+2*CO.6?

Yes.

Hmm, and the contact spacing and metal overlap DRC rules are subtracting 2*CO, to get correct results again? No, diffusion contact spacing rule is subtracting just one CO.6. Why just one?

The spacing between CONT shapes is 0.25um. The overlap of metal over CONT is 0.005um. So the minimum spacing between two contact shapes in magic should be (0.25 - 0.01 = 0.24), which it is.

Ah, so the description of that spacing rule is wrong.

Oh, I didn't notice the wrong description. Yes, it's just the description text that's wrong (or at least misleading). I can fix that.