Hi there, I'm running into the following error when simulating a synthesized design with ngspice (netlist attached)...

tom@tom:~/repos/projects/openwave$ ngs pll tb_pfd
setup_tcp_xschem: problems listening to TCP port: 2021
couldn't open socket: address already in use


Note: Compatibility modes selected: hs a

Warning: m=xx on .subckt line will override multiplier m hierarchy!


Circuit: 

Error: unknown subckt: x1.xfiller_13_144.x0 vdd x1.xfiller_13_144.a_124_375# x1.xfiller_13_144.a_36_472# vdd x1.xfiller_13_144.pfet_06v0 <http://x1.xfiller_13_144.ad|x1.xfiller_13_144.ad>=1.0736p x1.xfiller_13_144.pd=6.64u <http://x1.xfiller_13_144.as|x1.xfiller_13_144.as>=0.5368p <http://x1.xfiller_13_144.ps|x1.xfiller_13_144.ps>=3.32u x1.xfiller_13_144.w=1.22u l=1u
    Simulation interrupted due to error!

Note: No ".plot", ".print", or ".fourier" lines; no simulations run

I have the following included in my top level spice dec but it still seems to be unable to find the 6v fets as far as i can tell

.include "/home/tom/repos/edalibs/pdks/gf180mcu/gf180mcuC/libs.tech/ngspice/design.ngspice"
.lib "/home/tom/repos/edalibs/pdks/gf180mcu/gf180mcuC/libs.tech/ngspice/sm141064.ngspice typical
.lib "/home/tom/repos/edalibs/pdks/gf180mcu/gf180mcuC/libs.tech/ngspice/smbb000149.ngspice typical"

Anyone have any ideas? Thanks in advance, Tom

seems like it can't find

xfiller_13_144

I was able to simulate the netlist. I had to uncomment the .subckt pfd line and related .ends line to avoid an empty netlist (no instances). SImulation of course is meaningless since there are no stimuli but no errors.

@Tom: Like Stefan, I had no difficulty simulating this netlist (I created a top-level wrapper for it and ran a

.op

). The error message indicates that it cannot find a subcircuit for

pfet_06v0

. That should be in your included file

sm141064.ngspice

at line 47062:

.subckt pfet_06v0 d g s b w=1e-5 l=5e-7
+ as=0 ad=0 ps=0 pd=0 nrd=0 nrs=0 par=1 dtemp=0
+ sa=0 sb=0 nf=1 sd=0 m=1
...

What do you see in file

/home/tom/repos/edalibs/pdks/gf180mcu/gf180mcuC/libs.tech/ngspice/sm141064.ngspice

?

I seem to have different content in that area

47055 m0 d g s b nmos_6p0 w=w l=l as=as ad=ad ps=ps pd=pd nrd=nrd nrs=nrs
47056 +delvto='mis_vth*sw_stat_mismatch'  sa=sa sb=sb nf=nf sd=sd
47057 .ends nmos_6p0
47058 *------------------------------------------------------------------------
47059 .subckt pmos_6p0 d g s b w=1e-5 l=5e-7
47060 + as=0 ad=0 ps=0 pd=0 nrd=0 nrs=0 par=1 dtemp=0
47061 + sa=0 sb=0 nf=1 sd=0 m=1
47062 
47063 .param
47064 + par_vth=0.01051 
47065 + par_k=0.00517 
47066 + par_l=3e-7
47067 + par_w=-4e-7
47068 + par_leff='l-par_l'
47069 + par_weff='par*(w-par_w)'
47070 + p_sqrtarea='sqrt((par_leff)*(par_weff))'
47071

there is a

pmos_6p0

on a near by but different line number and no

pfep

. My open_pdks git hash is

commit e6f9c8876da77220403014b116761b0b2d79aab4 (HEAD -> master, tag: 1.0.393, origin/open_pdks-1.0, origin/master, origin/HEAD)

and openlane

commit ce916510de89a037889e06ab16c00c4101f77d2f (HEAD -> master, tag: 2023.02.08, origin/master, origin/HEAD)

The difference between the names

pfet

and

pmos

is what's causing the issue. The open_pdks version number is current, but that won't necessarily tell you what is the version of the

gf180mcu_fd_pr

library. Just before I tested the simulation myself, I deleted all the content of

open_pdks/sources/gf*

and ran "make" to let it re-clone the open PDK from scratch. Maybe you set up the

configure

for open_pdks pointing to a pre-existing (and outdated) version of the PDK?

I see. Ok I'll check that later tonight. I usually do a git clean fdx on open pdks before pulling followed by a configure, make, etc... Would that not have the same effect and pull in the latest repos?

Try

make distclean

to delete the cloned sources, or

make update

to do a git pull on each of them. open_pdks puts the cloned repositories into subdirectory sources/, but a git pull on open_pdks itself will not overwrite those, so an update of those PDKs has to be done manually or by

make update

from within open_pdks.

I think that may have been the problem

tom@tom:~/repos/edatools/open_pdks/sources/gf180mcu_fd_pr$ git log -1
commit 74e4ec59b55bcf5be2f153abff8519d15ebe21fa (grafted, HEAD -> main, origin/main, origin/HEAD)
Author: Tim 'mithro' Ansell <tansell@google.com>
Date:   Fri Nov 4 11:12:05 2022 -0700

    Merge pull request #50 from umarcor/umarcor/ci-klayout
    
    ci/Tests: bump Klayout to 0.27.10

looks like

git clean -fdx

may not have cleaned out the sources folder for some reason. running a

make distclean && make update && make install

now to see if that fixes the issue

@Tom: (1) You either do

make distclean

to get rid of the cloned repositories or

make update

to update them. You would not want to do both. (2) I think the underlying problem here is that we took the unusual step of changing the repository from Google to an Efabless fork (specifically the gf180mcu_fd_pr repository) because the Google repository was falling way behind our development. That is probably why the git update failed both for you and for the Makefile, because the entire repository changed. So

make distclean

was the only meaningful course of action in this case. At this point, I'm not sure what all the updating did to corrupt everything, but given that you already blew it all away and started clean, I'm going to assume that worked, and not worry about it too much.

Yes, I finally got it installed again, thank you @Tim Edwards. It took a while as for some reason I had some issues while downloading while running make with premature EOFs and closed connections. Not sure of the cause since my connection seemed very stable but it looks like I now have the correct model. I think I was massively out of date. Thanks for the help in getting it fixed