Hi <@U01819B63HP>, we are looking for the following xschem netlisting feature, not sure, if this is ...
h
Hi @Stefan Schippers, we are looking for the following xschem netlisting feature, not sure, if this is already existing. We would like to be able to netlist for LVS with certain elements removed (leaving the connecting wires open or short-circuit). Means, we need • a cmdline switch to indicate netlisting for LVS • an lvs_ignore property on symbols, which can be set to none/short/open (none doing nothing, short = remove element, short nodes, open = remove element, keep wires floating) Is this someting you could implement? That would be very helpful!
s
@Harald Pretl That is interesting. There is already something in place for this. I am looking at it and will come with a solution for that. There are some low level commands to do this, but I want an easy interface. WIll keep you updated.
@Harald Pretl are you assuming 2 terminal symbols (like -say- a metal resistor / metal option) so you have 3 options for the LVS netlist: • keep the resistor /option • short the 2 nodes together • remove the resistor /option If more than 2 terminal symbols are used what should I do ? i think about this: • keep the symbol in the netlist • short all nodes together • remove the symbol from netlisting Does that make sense?
h
I think we can restrict to simple cases, as mostly this will be used for elements like R/L/C (parasitic modelling) or sources (for stability simulation). So with mostly 2-terminal devices in mind, a simple remove-and-short or remove-and-keep-open should be sufficient.
The background is that having cell schematics both for simulating as well as LVS, without the need to modify them. At some point in the project, cells should be read-only, just to be sure 🙂
s
ok, thank you, will look at a simple solution for this.
👍 1
@Harald Pretl I have added the lvs_ignore switch, please see the video and let me know if that fits.
h
@Stefan Schippers Super! That is a perfect implementation! I’ll try it the next week on our design, first I need to put the latest xschem with this changes into our Docker image, but that looks really promising. BTW, the flag
top_subckt
is fully retired now? Means, I do now a
set lvs_netlist 1; set lvs_ignore 1
, correct?
s
Of course that needs some testing. I am doing that on my side too, so don't rush it into production immediately 🙂 (i mean the new features, i don't expect regressions due to this).
yes
top_subckt
is now renamed to
lvs_netlist
. I have kept
lvs_ignore
,
lvs_netlist
and
spiceprefix
all separate so user can combine them for his needs.
h
@Stefan Schippers The new
lvs_ignore
feature works like a charm 😄 I built the latest
xschem
and tested it on our SAR ADC, and now the LVS is passing w/o hand-manipulation of the netlist. Thanks!
s
Wow, that's nice. Thanks for testing that immediately