PEX Netlist Simulation in Spectre

spectre netlist simulation

Hi all,

Currently I'm doing post-simulation for my ring oscillator.
I have issue with the PEX Spectre netlist generated by Calibre.
I've tried 2 methods for simulating this PEX netlist.
1) ADE method:
I created a symbol to be instantiated into my test bench.
I referred to this site:
https://www.cadence.com/Community/b...a-subcircuit-netlist-with-spectre-in-ade.aspx
--> The output: Oscillation starts only at about 2.5ns.
The first few lines in the spectre.out:
V(N_net065_c_196_n) = 0 V
V(N_net065_c_199_n) = 0 V
V(N_net065_c_200_n) = 0 V
V(N_net065_c_201_n) = 0 V
V(N_net065_c_202_n) = 0 V
V(N_net065_MNM5_s) = 0 V
V(N_net065_MPM3_s) = 0 V
V(N_net065_RR3_pos) = 0 V

2) Terminal method:
I included all the model libraries inside one directory.
I've used the Linux terminal to simulate this pex netlist.
command: spectre xxx.scs
--> The output: Oscillation starts only at abt 90ns.
The first few lines in the log file:
V(N_net065_c_196_n) = 1.75626 V
V(N_net065_c_199_n) = 1.7564 V
V(N_net065_c_200_n) = 1.75675 V
V(N_net065_c_201_n) = 1.75631 V
V(N_net065_c_202_n) = 1.75631 V
V(N_net065_MNM5_s) = 1.75631 V
V(N_net065_MPM3_s) = 1.75626 V
V(N_net065_RR3_pos) = 1.75681 V

So, my point of having these 2 methods is to compare the output, because if I run HspiceD simulation for the PEX HspiceD netlist generated by Calibre -> I don't have this kind of late oscillation!
Note that the simulation settings are similar.

Anyone can explain why?

Thanks a lot!