@Faedra Webers Also try to change integration method, may be to backward euler (
.options Method=trap maxord=1
).
Also increase ITL4 (ITL4=200).
"_MAXORD=x [*] specifies the maximum order for the numerical integration method_
used by SPICE. Possible values for the Gear method are from 2 (the default) to 6.Using the value 1 with the trapezoidal method specifies backward Euler integration.METHOD=name sets the numerical integration method used by SPICE. Possiblenames are ‘Gear’ or ‘trapezoidal’ (or just ‘trap’). The default is trapezoidal.ITL4=x resets the transient analysis time-point iteration limit. The default is 10.
_CHGTOL=x resets the charge tolerance of the program. The default value is 1.0e-14._"
👀 1
f
Faedra Webers
08/15/2022, 10:39 AM
Thank you for your reply!
I tried your settings, and the result stayed the same. I tried to increase them even further ITL4= 1000 and CHGTOL = 10e-18 , but simulation is still running after 2 hours...
s
Stefan Schippers
08/15/2022, 9:35 PM
@Faedra Webers another thing to test is the .tran timestep.
The tran (or .tran) command accept up to 4 parameters:
.tran <tstep><tstop> [<tstart> <tmax>]
you can try to lower tstep and tmax (and reset CHGTOL and ITL4 to less aggressive values)
Stefan Schippers
08/15/2022, 9:38 PM
If you still see nodes diverging may be there is a off-state leakage problem? try to increase the channel length of transistors or replace _lvt with standard transistors.
Linen is a search-engine friendly community platform. We offer integrations with existing Slack/Discord communities and make those conversations Google-searchable.