This is the issue actually: for some reason, the PMOS wrapper generated by the block netlist identif...

Jorge Marin

12/31/2021, 1:04 PM

This is the issue actually: for some reason, the PMOS wrapper generated by the block netlist identifies the bulk node, but this is not the case for the NMOS wrappers generated, as seen below (the PMOS wrapper has 4 pins while the NMOS wrapper has 3 pins):

.subckt sky130_fd_pr__pfet_01v8_MA8JHN *a_15_n136#*

a_n33_95#
a_n73_n136# w_n109_n198#

X0 a_15_n136# a_n33_95# a_n73_n136# w_n109_n198# sky130_fd_pr__pfet_01v8 ad=0p pd=0u as=0p ps=0u w=1e+06u l=150000u

.ends

.subckt sky130_fd_pr__nfet_01v8_59MFY5 *a_n73_n69# a_n33_n157# a_15_n69#*

X0 a_15_n69# a_n33_n157# a_n73_n69# VSUBS sky130_fd_pr__nfet_01v8 ad=0p pd=0u as=0p ps=0u w=1e+06u l=150000u

.ends

.subckt sky130_fd_pr__nfet_01v8_6H9P4D *a_n73_n100# a_15_n100# a_n15_n126#*

X0 a_15_n100# a_n15_n126# a_n73_n100# VSUBS sky130_fd_pr__nfet_01v8 ad=0p pd=0u as=0p ps=0u w=1e+06u l=150000u

.ends

Luis Henrique Rodovalho

12/31/2021, 1:14 PM

The fourth NMOS terminal is the substrate, VSUB, and it isn't showing in the subcircuit. The PMOS wrapper doesn't count it VSUB, as its bulk terminal, the nwell connection, is explicit.

Jorge Marin

12/31/2021, 1:37 PM

OK, I see... the question is how to force the NMOS wrapper to have the fourth terminal, so that the LVS doesn't detect each bulk node as a different net I don't know how to do this since these subcircuits are generated automatically

Luis Henrique Rodovalho

12/31/2021, 1:41 PM

This is an issue with "ext2spice hierarchy off", isn't it? If the parent cell has both PMOS and NMOS devices, extraction should work and both single device subcircuits should have their four pins.

Tim Edwards

12/31/2021, 4:11 PM

@User: Update your version of

magic

. There are some recent bug fixes that could be related. In particular, there were some fixes around nets which were not showing up in the pin list. What version of magic are you using?

Jorge Marin

12/31/2021, 6:09 PM

hello @User, I'm still using magic v8.3.230

Tim Edwards

12/31/2021, 6:21 PM

Very likely that this issue was solved in rev 235 or 241. Also revs 244 and 245 are related, but probably not what you're seeing in the output.

Jorge Marin

12/31/2021, 10:28 PM

Thanks for the reply, I've accumulated various issues and I won't be able to finish before the deadline, so unless you extend it few more days it will be for MPW5 (hopinng there will such run!)

Open in Slack

Previous Next

What is Linen?

Linen is a search-engine friendly community platform. We offer integrations with existing Slack/Discord communities and make those conversations Google-searchable.