Gmindc different than the value set in the Options/analog

This is explained by Spectre can now adjust gmin value over a range automatically to resolve convergence issues . If you set gmindcmax you can limit the upper limit that gmindc will be adjust to in the case of "no unique solution".

You can also just make the "Circuit has no unique solution" an error, following How to change spectre warning messages into error messages when Spectre determines that the circuit has no unique solution

Andrew

Dear Andrew,

Thanks a lot for your explanation and help. Unfortunately, the issue is still present. I set gmin=1e-14, gmindc=1e-14 and gmindcmax=1e-12.
However, the simulator still uses Gmindc=100 uS. Any other idea?

*****************************************************************

Global user options:
psfversion = 1.4.0
vabstol = 1e-06
iabstol = 1e-12
temp = -40
try_fast_op = no
gmin = 1e-14
gmin_check = all
gmindc = 1e-14
rforce = 1
maxnotes = 5
maxwarns = 5
digits = 5
cols = 80
dc_pivot_check = yes
pivrel = 0.001
sensfile = ../psf/sens.output
checklimitdest = sqldb
gmindcmax = 1e-12
save = selected
currents = selected
reltol = 0.0001
tnom = 27
scalem = 1
scale = 1

Scoped user options:

Circuit inventory:
nodes 5481
bsim4 916
bsource_2ab5b1 258
bsource_09f256 2800
bsource_79a729 416
bsource_7533b7 426
bsource_bd39ed 1400
bsource_c39284 1613
bsource_rpldd_50k_noi_225128 60
bsource_rpposab_8ef6d6 48
capacitor 460
dac_8bit_ideal 1
diode 1933
iprobe 6
isource 1
resistor 1613
vccs 1
vdc8bit 1
vsource 15

Analysis and control statement inventory:
info 6
stb 1

Output statements:
.probe 0
.measure 0
save 0

Design checks inventory:
paramtest 1
assert 176

Notice from spectre during initial setup.
Protected devices exist and are not included in the circuit inventory.
365 warnings suppressed.

Time for parsing: CPU = 205.993 ms, elapsed = 491.271 ms.
Time accumulated: CPU = 709.027 ms, elapsed = 1.08937 s.
Peak resident memory used = 181 Mbytes.

~~~~~~~~~~~~~~~~~~~~~~
Pre-Simulation Summary
~~~~~~~~~~~~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~~~~~

**************************************************
Stability Analysis `stb': freq = (1 Hz -> 100 MHz)
**************************************************
Matrix is singular (detected at `I_DUT.I_MIRR_2_2.I_OPAs.D4.d0:int_c').

Warning from spectre during DC analysis, during STB analysis `stb'.
WARNING (SPECTRE-16085): Circuit has no unique solution.
A resistance of 10 kOhm (1/GminDC) was added across each nonlinear device.
Notice from spectre during DC analysis, during STB analysis `stb'.
GminDC = 100 uS is large enough to noticeably affect the DC solution.

First I would suggest checking with the latest SPECTRE25.1 ISR and if the issue persists, then please contact customer support (after logging in, use the case menu to submit a support case). We'll probably need your example netlist that shows the problem.

Andrew

First I would suggest checking with the latest SPECTRE25.1 ISR and if the issue persists, then please contact customer support (after logging in, use the case menu to submit a support case). We'll probably need your example netlist that shows the problem.

Andrew

Understood. Thanks for your support