IIP3 simulation using transient analysis

Has anyone tried IIP3 simulation using transient? For example if my
two input tones are @ 2.45G and 2.451G and my LO signal is @ 2.44G
(using PORT for all signals). How long do I need to run transient for?
How do I choose the time step? And how should I specify the range and
number of samples when calculating DFT?

I tried simulating but when I take the DFT @ the output of the LNA I
can only see one tone, I can never see 2 fundamental and the IM3 tones.
I think it has to do with resolution but I am not sure how to solve it.
Any help will be appreciated.

Thanks

 

Your fundamental frequency should be 1MHz, so the simulation time
should be x ns to x + 1000 ns, where x is the amount of time
required to settle to sinuosoidal steady-state. Usually circuits
and simulators require a little time to settle start-up transients.
One other option if you are using Spectre, is to use Spectre's
Fourier Analysis;

four out 0 fourier fund=1MHz harmsvec=[ 5 6 7 8] normharm=6

This will give you the magnitudes for the fundamentals and third-order
distortion products. If you try an FFT, it probably should be more
than 8192 points to prevent the LO leakage from aliasing.