I am trying to setup a bench to check the transfer gain from VDD, GND and Substrate to Oscillator output. My oscillator is a PMOS cross coupled LC Osc. Here is what I do:
Test1: PSS+PXF
PSS form:
PSS Analysis: Shooting
Beat frequency = 4G
Output frequency = 10
Oscillator node positive = osc_out
Oscillator node negative = gnd!
PXF form:
Sweeptype = relative harmonic =1
Start=1K, Stop=1M
MaxSidebands =10
Oscillator node positive = osc_out
Oscillator node negative = gnd!
Test2:
PSS+PAC
On schematic:
  PAC=1 is set on the vdc source connecting the node (VDD/ GND/ Sub).
PSS form: Same as above
PAC form:
Sweeptype = relative harmonic =1
Start=1K, Stop=1M
MaxSidebands =10

I am trying to find how the noise from the low frequency gets mixed upto the oscillating frequency.
For PXF simulation output: should I check the voltage gain of sideband given by 1K-1M or 4G+1K to 4G+1M?
For PAC simulation output:
  Why does Voltage gain have input and output sidebands while voltage only has output sideband?
  Should I check the voltage for sideband given by 1K-1M or 4G+1K to 4G+1M?
I want to make sure that both PXF and PAC are giving identical results (accuracy issues ignored) and that I am actually plotting the frequencies of interest.

Always been confused. I have read the 'Introductionto RF simulations and its application' and still couldn't get the spectre usage of it.