AMS Convergence Problem

Dear All,

I am facing a convergence problem. The parameters I have set for my
simulation are as follow:

start = 0 s
outputstart = 0 s
stop = 1.5 ms
step = 1.5 us
maxstep = 30 us
ic = all
skipdc = no
reltol = 10e-03
abstol(V) = 100 uV
abstol(I) = 100 pA
temp = 25 C
tnom = 27 C
tempeffects = all
errpreset = moderate
method = trap
lteratio = 3.5
relref = sigglobal
cmin = 100 aF
gmin = 1 pS

But whenever I am trying to run the simulation, the simulator says
that "Trying `homotopy = dptran' for initial conditions.". But after
that nothing happens after waiting a long time. I have tweaked several
parameters like integration methods etc but no result obtained.

Can any one please help me in this regard?


Thanks in advance for help.

Regards !!!!!

Debjit.

Cheers !!!!!!

 

Debjit,

You are using this forum for quite a while and you should know by now
that the information you have providedis not enough for us to give you
any help.
How would I proceed in you place ?

1. I will tell the community what I am simulating, i.e. ADC, VCO,
PLL ... etc.
2. I will tell the community whether my full Spectre simulation
(outside the AMS env) was fine.
3. I will tell the community whether my full Spectre simulation within
AMS, i.e. all cell in the HED are tied to a spectre Subckt (not any
HDL models) is giving the same results as 2.
4. I will tell the community my tool versions, i.e. icfb -W, spectre -
W and nchelp -v
5. I will tell the community whether I use AMS in a ADE use model or
HED-plugin use model
6. I will tell the community whether I run the 3-steps CellBased
netlisting flow or the 1-step OSS irun flow ...

Please give us some more details if you want any further followup ...

Regards,
Riad.

 

Hi Riad,

I am extremely Sorry. I beg pardon humbly for not providing enough
info.
The details are given below.

1. Circuit Type: PMU Chip consisting of LDOs and Buck Converters
2. Simulation Result: Suceessfully simulated in Circuit Level using
Spectre and with Verilog-A and Verilog models using spectreVerilog.
3. Basically whenever I am trying to circuit in AMS (as I am gradually
switching from spectreVerilog to AMS following the AVUM workshop
manual included in the Cadence Distributino), it says that it is
trying to do homotopy = "dptran" but after some hour, it reports that
it failed to converge. I obtained convergence using homotopy =
"dptran" in spectreVerilog.
4. icfb Subversions: 5.10.41.500.6.141
spectre Subversion: 7.2.0.174.isr3
ncsim version: 08.20-s022
nchelp version: 08.20-s022
5. AMS used in ADE model. I don't feel comfortable in HED model.
6. I am using OSS irun based netlister and simulating in batch mode.
7. I am using Cadence in built connect modules.
8. The summary of simulation parameters are given as follows:
start = 0 s
outputstart = 0 s
stop = 1.5 ms
step = 1.5 us
maxstep = 30 us
ic = all
skipdc = no
reltol = 10e-03
abstol(V) = 100 uV
abstol(I) = 100 pA
temp = 25 C
tnom = 27 C
tempeffects = all
errpreset = moderate
method = trap
lteratio = 3.5
relref = sigglobal
cmin = 100 aF
gmin = 1 pS
The above simulation setup was fine for me in spectreVerilog.

Thats all the info I have with me.

Please provide me some suggestions.

Thanks and Regards !!!!
Cheers !!!


Debjit.