I want to try to make this as comprehensive as possible. I have read and tried following the sc filter and delta-sigma spectre pdf's.  

It is hard for me to correlate the results I get in spectre with the documents because the documents do not show actual spectre results.

The circuit I am working with is a sc integrator with chopping. I have ran the simulation 2 times(chopping on, chopping off). The noise results are very close to identical with the chopping noise numbers higher.

As you can see I implemented the ideal S/H from the sc filter pdf into my feedback. I created extra output ports to be able to track and monitor when the input is being read in and when the output is be updated. the timings seem correct in the spectre pss trans sim
(p1= sample phase, p2= int phase, ch8/b = chopping clocks, pout= time the feedback SH updates its output)
in the noise plots vac(op,on) = the pac results at the output with respect to 1 volt ac input, gain_var is a transformed periodic gain result with all of the foldings implemented back into base band, n() is the noise at the output, and n()[0]/gain_var is my attempt at input referred noise. I realize that spectre has iprobe statements that should record input referred noise, but my company's software wrapper seems to not save the iprobe results.

My question: why do I not see reduction of 1/f noise in the chopped version, and when I change the feedback caps around the integrator to be a gain of 2, I do not see a gain change in the pac results.

My circuit is integrating at 768k, the chopping is occurring at 384k.

pnoise settings:

smpldNoise (op,on) pnoise iprobe=V19
+        annotate=status
+        maxsideband=100
+        start=0.01
+        stop=10e8
+        dec=40
+       relharmnum=1
+       refsideband=1
+        noisetype=timedomain
+        noisetimepoints=[1.3u]
+        numberofpoints=0

Here are the schematic and result screenshots.




No Chopping

Chopping

Please help, These noise numbers are holding back our design schedule
-Pb

That was attempted a wile back. We havent been impressed with our regional cadence rep when it comes to switch cap noise simulation. he seems to be more geared for rf and oscillator simulation type questions. But I will retry the attempt. thank you

ps - A coworker in a different group/location told me that while noisetype=timedomain is intended for sampled systems, that if i run the output through an ideal S/H and run pnoise on that output that i can just use the noisetype=sources and pac sidebands of 0, since the S/H does the folding in its own transfer function. This also lets me use n()/pac() without having to transform the pac results.  Have an opinion on this?