I read the <<simulating switched-capacitor filters with spectreRF>> by ken. I try to follow it. And I have simulated a switch capacitor filter. In the simulation, I found that the noise is double when the maxsideband and the number of harmonics changed from 100 to 200. I don't think the harmonics larger than 100 will contribute so much noise. So the question is , what's wrong with it? One part of netlist is listed below.

pss  pss  fund=1M  harms=200  errpreset=moderate  tstab=4.6u
+    method=gear2only  tstabmethod=gear2only  annotate=status  maxiters=50
pnoise  (  outp  outn  )  pnoise  start=1  stop=1e6  dec=101
+       maxsideband=200  iprobe=V6  refsideband=0  annotate=status

I also include the noise file in the vdd. The frequency range is 10-20kHz. The total output noise is also measured in this range. The another question is , if I only provide 10-20kHz range noise, how will the simulator deal the noise out range of it? Is the noise file used in this method right or not?

thanks.