Pranav Lulu

04/17/2023, 12:59 PM
@Tim Edwards @Mitch Bailey With the latest magic version
I am getting some drc errors which apparently are not flagged by klayout. Here's the drc report and the gds for same.
The Deep N-well spacing to N-well is a known issue in the magic I guess which was reported earlier.

Mitch Bailey

04/18/2023, 3:30 AM
I don’t have much to do with drc. Sorry, I couldn’t help. You might ask @Amro Tork about the klayout rules. I think he has experience with those.

Tim Edwards

04/22/2023, 2:08 AM
@Pranav Lulu: Sorry for the late reply, as I have been in Europe. This layout was not done in magic, and like many designs done in other layout editors, devices and their constituent layers have been divided up between subcells; a very bad practice which other tools unfortunately encourage (since extraction tools like Calibre are capable of going through immensely complicated hoops to automatically restructure the layout to make it extractable). In this case (I think---I haven't stared at it too closely) the HVI layer is split between different cells which is causing some areas to read in as low-voltage diffusion and others as high-voltage diffusion, in the same general area, causing high voltage-to-low voltage diffusion spacing errors where there are none. Sometimes it is not possible for magic to read in such a layout in a way that will render it DRC clean. Sometimes it can be made DRC clean by selectively flattening specific cells. Since this layout is apparently from Cadence and has Cadence-specific pcell references using "$$", you might try selectively flattening all cellnames containing the double-dollar sign using the
gds flatglob