gf180mcu magic tech file does not connect salicided diffusion with different implants.

[d-m-bailey]

The gf180mcu process uses salicide to reduce the resistance of diffusion and poly.

This has the consequence of causing abutting diffusion with different implants (N+ and P+) to be electrically shorted. Without salicide, this junction would form a diode, but with salicide they are the same potential.

The Chipathon2025 had a design with abutting diffusion of different potentials that was not detected by magic extraction.

One suggestion is to change

connect
  nwell,*nsd,*mvnsd,nbase,dnwell nwell,*nsd,*mvnsd,nbase,dnwell
  pwell,*psd,*mvpsd,pbase,isosub  pwell,*psd,*mvpsd,pbase,isosub
  *psd,*mvpsd  *psd,*mvpsd
  *m1	*m1

to

connect
  nwell,*nsd,*mvnsd,nbase,dnwell nwell,*nsd,*mvnsd,nbase,dnwell
  pwell,*psd,*mvpsd,pbase,isosub  pwell,*psd,*mvpsd,pbase,isosub
  *psd,*mvpsd  *psd,*mvpsd
  *nsd,*mvnsd,*psd,*mvpsd  *nsd,*mvnsd,*psd,*mvpsd
  *m1	*m1

Unrelated question:
Isn't the connectivity in

  *psd,*mvpsd  *psd,*mvpsd

included in this line?

  pwell,*psd,*mvpsd,pbase,isosub  pwell,*psd,*mvpsd,pbase,isosub

[d-m-bailey]

@RTimothyEdwards the suggestion above is insufficient. It only connect taps. To connect taps and source/drains, this is probably needed.

  *nsd,*mvnsd,*ndiff,*mvndiff,*psd,*mvpsd,*pdiff,*mvpdiff  *nsd,*mvnsd,*ndiff,*mvndiff,*psd,*mvpsd,*pdiff,*mvpdiff

[RTimothyEdwards]

I added the correction, which will appear along with the rest of today's edits in open_pdks version 1.0.555.