Hi guys, I'm recently confusing running my post-simulation with parasitic extracted. Two ways are used to create the circuit to be test from the same layout, but I get different simulation results.
All parasitic is extracted by Calibre PEX.
First way. Calibre PEX-->Outputs-->CALIBREVIEW (C+CC reduction disabled) A calibre view with parasitic was created in my circuit cell. Running the testbench with calibre view after hierarchy edited.
Second way. Calibre PEX-->Outputs-->SPECTRE (C+CC reduction disabled) An "xxx.pex.netlist" file was created. A spectre view was copied from symbol view of my circuit. Then edit the spectre view through CIW-->Tools-->CDF-->Edit. Terminal order edited. The netlist file was included in model library in testbench. Running the testbench with spectre view.
However, the simulation results is really different in two ways. Green the spectre and red the calibre, different amplitude applied in testbench and no transient noise. So which way is trustworthy? And why comes the difference since the netlist is all extracted by the same PEX process? Theoretically, the netlists of the two circuits with parasitics should be the same, so as the simulation results.
This is something you should ask Siemens EDA since both views/files are being output by Calibre, which is their tool (not a Cadence tool). Presumably there must be some difference in the content to give different results - so I suggest you ask them.
Thanks Andrew, it's really like something wrong in Calibre. A trivial parasitic capacitor of several fF collapsed the circuit, and that can't be explained by circuit theory.