So, Cakewalk originally designed the Console Emulator to utilize the Console Emulator "Bus version" on the Master Bus. OK, it's all about the cumulative effect. Then, Craig suggested in an earlier thread that he did not use any emulator on the master bus.
 
Now, we're shown to use the "Channel" Emulator on the Master Bus.
I'm trying to follow these leads, but my ears tell me it's very subtle (unless you crank the outputs)
 
So, I guess what I'm getting from this is to follow your ears and experiment, but I'm not sure I have that kind of time to dedicate to experimentation. I have Console emulator project template presets that I'm using less and less over time, because they add an unpredictable warmth and low + mid range content that I have to compensate in my mixes elsewhere. Craig, your tips are always valued here . . . but I'm just not getting this one.

I've been using the CE this way since it came out.  I like the ability to adjust both the Trim and Drive to taste, so I never really used the Buss version.  

@noynekker The reason I used the channel CE on the bus was so I could bump up the trim, and therefore add some of the transformer saturation the CE emulates.
 
As to Mister Happy's test, yes, the CE as shown in the tip will add level (I saw less than 2.5 dB, but whatever - maybe the controls were turned up a bit higher). However, if you think the only differences are from the level increase then you're missing what the CE is adding.
 
Although I greatly appreciate anyone making the effort to try and come up with a test to determine whether the tip has merit or not (and I wish more people would try things out for themselves!), unfortunately the test is flawed in three fundamental ways so it's impossible to draw any meaningful conclusions - see the screen shot showing spectral energy and correlation.
 

 
1. The material is pretty much mono, so the CE can't affect the stereo imaging because there isn't really any stereo imaging to affect.
 
2. Compared to most mixes, the highs are way down compared to the lower frequencies. Like imaging, the CE influences the high frequencies but there aren't enough highs for the CE to "chew on."
 
3. Most importantly, the source material isn't the same for the two of them. Do a null test, and you'll hear that one appears to have reverb or other ambience, while the other doesn't.
 
As a result, the test is meaningless. For a more meaningful test, I took a short clip of music, copied it, processed one with the settings shown in the tip and rendered it, then normalized both clips to -0.1 dB. I then combined them to a single clip that you can listen to at the following URL.
 
https://1drv.ms/u/s!Am5TdLYuWxQOkhw2n1E5Mj5BWRSm
 
I think the difference is subtle, but nonetheless pretty obvious. If you don't think so, listen carefully to the difference in the snare's attack transient, the cymbal at the end, the imaging for the guitar sound, and the high-hat that appears around the middle of each clip. I'm not only quite sure you'll hear a difference, but will prefer one of the clips over the other. That's the one with the CE.
 
 

.

.

mister happy
...However the “pairs” were exports identical in every way except the CE on and off. The null test shows what the CE creates as a difference while the ABX test shows how often that difference may be perceived.

Are you sure one of them wasn't going to a reverb bus? If you do a null test with the two clips I posted, there is nothing remotely like the artifacts on the two files you posted. I don't think there's any way the CE can isolate ambience, but hey...maybe it's what happens with (mostly) mono files. 

.

mister happy
Oh, and I would not suggest normalizing to peak but rather match to the average RMS or perhaps the LUFS levels.

Yes, but Sonar doesn't have the option to measure those so it wouldn't help those who wanted to check things for themselves. I checked the LUFS in Studio One for the two clips and they were really close.
 
A quick and dirty way to A/B test in Sonar is to create two tracks, get everything matched up properly, and group each track's solo buttons when one solo is on and the other off. Then you can hit either solo button repeatedly with your eyes closed until you don't know which is which, and then compare the two sounds without being influenced by what you see.
 
The bottom line for me is the CE is subtle when used as intended, but it can make a positive difference. And if you turn the Trim control up all the way on bass tracks, it's really cool 

Anderton

mister happy
Oh, and I would not suggest normalizing to peak but rather match to the average RMS or perhaps the LUFS levels.

Yes, but Sonar doesn't have the option to measure those so it wouldn't help those who wanted to check things for themselves. I checked the LUFS in Studio One for the two clips and they were really close.
 
A quick and dirty way to A/B test in Sonar is to create two tracks, get everything matched up properly, and group each track's solo buttons when one solo is on and the other off. Then you can hit either solo button repeatedly with your eyes closed until you don't know which is which, and then compare the two sounds without being influenced by what you see.
 
The bottom line for me is the CE is subtle when used as intended, but it can make a positive difference. And if you turn the Trim control up all the way on bass tracks, it's really cool 

Oh cool, I need to try that on Bass!

.