@User Looking at the gds for one of the pico designs https://efabless.com/pprojects/560, it seems that floating nsdm and psdm implants do not flag errors even when there is an overlap at a higher level. In the cells where these implants are placed (

nmos20

,

pmos20

) there is no other layer. So on cifinput, it appears magic totally ignores these shapes and thus no drc errors, correct? I hear there were no klayout errors either (passed precheck).

We were never sure why that project was so screwed up. We attempted to get in touch with the designer to get it corrected. It did not pass precheck. I don't think we heard back from the designer. Magic does not read implant layers because it will erase them and write out its own implant layers. If it reads in GDS as "readonly", then it assumes it is DRC correct. To catch such DRC errors in magic, you need a different tech file. There is a

sky130A-GDS.tech

file that will catch these errors, but the precheck uses klayout since that checks the same errors plus a lot of other ones as well.

Thanks for the explanation. Glad to hear that it wasn't a major hole in verification. This project also has a couple connections via substrate. I've reported them to Boris Murmann from SSCS and Muhammad Rana of FAST National University.

Hi @User, I was one of the designers on the chip, though I did not design that specific module. The project did pass precheck on the efabless site. Can you please elaborate? It will help to avoid these issues in future.

We were going through a revision of the "manufacturing rules" given to us by SkyWater at the time, so it's possible that it passed precheck at the time you submitted it. SkyWater rejected the reticle after "visual inspection" which led us to point out to them that these rules were not in their "manufacturing rules" DRC deck. We then tried to get in touch with the designers of the several chips that were causing the reticle to be rejected, and ultimately had to remove those projects that we got no answer or resolution from in time, so that we could go ahead with the shuttle run.