If you are working with a receiver, are you computing a transfer function from an RF input to a baseband or IF output? If so, are you using an RF simulator like SpectreRF, EldoRF, or Golden Gate?

If you are looking at part of the receiver that does involve frequency conversion, the first thing that comes to mind is saturation. A transient analysis would account for saturation. Whether an AC analysis accounted for saturation would depend on the DC operating point and even so, it would not necessarily give the same result as a transient analysis.

-Jess

Does your receiver have a mixer driven by an LO? And are you trying to compute the transfer function from the RF input to the baseband or IF output?

OR, are you trying to compute the transfer function of across part of the receiver that does not include the mixer?

-Jess

Let's get back to your original question:

Why don't your AC and transient analyses of the agree on the transfer function from RF input to baseband/IF output?

Assuming I paraphrased your question correctly, there is no way your AC and transient analysis should give the same results. Your transient analysis is probably OK but your AC analysis is meaningless because for AC analysis, Spectre linearizes the circuit about a DC operating point. The linearization locks the LO to whatever voltage it has at time zero and leaves it there. Your circuit is time varying but your AC analysis assumes the circuit is time-invariant; Your AC analysis is performed about a DC operating point that does not exist because the LO is not a DC signal.

Please forgive me if I have misunderstood your question.

-Jess

I guess I don't fully understand your question but it seems like you are wondering why you see different results with different methods of measuring the gain of just your LNA. To determine which method is right and which one needs work, I suggest replacing your LNA with a behavioral model. You can find such models in the rfLib or write your own in VerilogA. You can even use ideal voltage-controlled-voltage sources. With a behavioral model, you can specify the gain and match the terminals perfectly. Since you know exactly what gain you have, when you try the different methods of simulation you can quickly spot which one needs work.

-Jess