Here's a good way to get a handle on what the CE does.
I created three sine wave test tones - 100Hz, 700Hz, and 7kHz and loaded them into individual tracks in Sonar, then placed the CE channel version in the master. The tracks were set up for exclusive solo so I could A/B/C among them, and did my variations within the CE insedrted in the Master bus. The CE Trim was set to 1 o'clock, and the Drive up full to make the changes obvious.
The Trim button seems like a "more" or "less" control. If we accept that the CE adds non-linearities characteristic of analog electronics and transformers, then the Trim control simply choose how much you want to accentuate those non-linearities. The Drive control determines the nature of the non-linearities.
It's VERY easy to hear what happens to the waveform when you're listening to sine waves. The S is the most subtle and rounded, the N does the most coloring, and the A lies in between. I applied the CE to three 100Hz sine waves and loaded the waves into WaveLab to see the waveforms. The "look" of the waveforms correlated to what I was hearing: S was the most sine-like but the trailing wave of the positive peak is very slightly more of a line than a curve. A flattened the peaks more, but remained fairly rounded while N didn't flatten the peaks as much, but altered the waveform shape more radically.
With the low frequencies isolated, it sounded to me like these differences are intended to emulate the midrange "ring" you hear with transformers in the signal path.
I'll be covering this in depth in an upcoming Sonar Technique column in Sound on Sound.