I think most of this stuff is snake oil, so when the Console Emulator came out, I ran some tests on it. The CE is not only about creating the non-linearities inherent in the electronics of analog gear. I believe the distortion to which people allude is in large part due to input transformer emulation. Like tubes, audio transformers are processors; in particular, they generate distortion at lower frequencies.
 
Wendy Carlos jumped on digital recording early on, I believe it was an Akai MG1212. When I visited her studio, her MG1212 sounded so much better than other ones. I asked why and she explained that she'd added audio transformers on the I/O to improve the sound. Some of the "warmth" of tube mixers has less to do with tubes than the inductor-based midrange "ring." 
 
If you turn the CE Drive way up, it's like saturating an inductor. Although people generally don't use the CE as an effect, try taking the S-Type, putting it on electric bass, turning up the Drive control all the way, and hitting the input hard. You'll thank me 
 
I've done a lot of testing of the  ProChannel modules. There's some powerful stuff in there, but you need to know what everything does (e.g., how the four Quadcurve EQ curves affect the sound) to take full advantage of what it offers.
 
 

Thanks C.A.!
Good descriptions.

Maybe the best way to explain it is to actually think about older, analog desks and their characteristics. For a start, channels were matched across the desk but that matching could never be perfect. So each channel had slight variations. Engineers would often have a favorite channel or bus for certain tasks just because the result from that channel was better than the other channels. So this is where the tolerance switch comes in. It adds some subtle variation between channels/busses.
 
Craig mentioned output transformers in reference to drive. I'd agree. Analogue mixing desks, preamps etc. used valves or transformers on the output stages. Again, being analog this lead to imperfections in sound and colored the sound somewhat. Just changing the type of metal core used in a transformer would alter the sound to a very noticeable extent. It's this valve/transformer stage which is the biggest ingredient to the character of that "analog" sound. As well as each channel having different characteristics, each individual channel had different characteristics across its range. As such, each channel had a sweet spot and depending on the source and desired result, engineers would drive the channel slightly to hit that sweet spot. This would push the valve or transformer harder causing some subtle distortion and saturation. That seems to be what the drive knob is all about. You could also literally push the fader up to drive the sound on analog desks. There's a band I used to work with a lot in the early 90's who did a lot of acid jazz type stuff and had a phenomenal drummer. I used to "push" the snare, kick and toms fairly hard on the desk and onto tape (everything in the red - which you could do back then if you were a rebel) which gave them a lovely massive, phat, killer sound. To do the same kind of thing in sonar you'd probably need to push that drive up high and add some thick tape saturation. It would probably take a bit more than that though to really emulate the sound of driving into the red - in a pleasant way, but it could be done. So you can get creative with console emulators and tape emulators too.
 
It would be pointless using a console emulation on just one channel. You would be pushed to tell any difference. It will make a difference though when you use it on all channels and busses so it gives the effect of a console across the board (excuse the pun). The same goes for tape emulation. They are all accumulative effects that will give you similar results to if you were actually using an analog desk and tape with everything coming off tape through the channels of the desk and into the busses.

Good descriptions 35mm. Thanks.

Anyone have any info on the randomization of the tolerance control? Does it actually randomize or is it like off is 1 type 2 is another without randomization?

RSMCGUITAR
Anyone have any info on the randomization of the tolerance control? Does it actually randomize or is it like off is 1 type 2 is another without randomization?

I would imagine it uses some random seed to change the character slightly as randomising functions in software are simple and commonplace - probably easier than doing an offset.