different results from nanosim and hspice

[leebluer]

set_sim_eou

i have simulated an analog circuit by Hspice, and got satisfied results. then i want to do a power analysis by Nanosim. but i meet some problem that the transient result from Nanosim is different from Hspice.
it is the first time that i using Nanosim to do some analysis. i refer to the manuel file. but i still know nothing about the reason. could you give me some advice.
thanks

 

[dumbfrog]

nanosim set_sim_eou

there are different accuracy levels for nanosim, speed and accuracy tradeoff. u have to set the accuracy level to the max level, will be slow, inorder to have similar result as hspice.

 

[leebluer]

nanosim accuracy

Hi dumbfrog, thank you for your reply. I have set the accuracy of Nanosim to the highest level, but also got the different results from Hspice. so I am very confused about my design. what's wrong with it?

 

[santhoshv78]

nanosim accurate setting

nanosim is ment for fast simulation but not for accurate simulations.
You can beleive the result of Hspice than the Nanosim.
There are some commands that you can give in the configuration file that increases the accuracy of simulation.
Try adding these commands in the configuration file.

set_sim_eou sim=4 model=4
set_sim_tres 1n
set_print_vres 10u

 

[aidenbu]

nanosim　set_sim_eou

I have the same problem when I use nanosim to simulate the circuit. I am not very sure that if order "set_time_tres 1n" can increase the accuracy of the nanosim. For nanosim the default set of the time step is 10ps which is smaller than 1ns. Can such a set increase the accuracy of the simulation?

 

[dumbfrog]

using nanosim with hspice

try to ramp the power up from 0 to 100ns (0 to vdd) and check your resutls after another 100ns. some analog blocks take some time to settle down. if your circuit is a close loop simulation, it might take some time.

 

[duncan94022]

leebluer,

Nanosim and HSPICE use different simulation algorithms and you will not get the results to match. Use HSPICE for accuracy and only use Nanosim when the circuit is too large for HSPICE.

 

[leebluer]

hello duncan,

I use Hspice to design my circuit, and get satisfied results. I want to get some other information from Nanosim, such as dc path check, power analysis.
But I got different results between them. then should I believe the other information from Nanosim, for instance some about the power analysis?
I have set the accuracy to the highest level, but got nothing. The main problem is that the output voltage of bandgap is abnormal. At first, Nanosim cannot recognize some parameters of PNP model. After some changes(only to change the format, but not the value), I got the results, but it is much different from that from Hspice.

Thanks