Thanks Jeff - let me wrap my ahead around this info and post back. It's about bed time... :)

Hi Jeff,
 
NOTE: I am truly trying to understand the differences and appreciate the push-back, so I hope I don't appear (too) defensive! :)
 
I'm not sure I get the point of doing the panning/trimming. Even with a Dangerous 2 Bus you send it your stereo (and mono if preferred) tracks and it sums them together - exactly what I'm doing here. On something like the Dangerous 2 Bus, you can choose to have a pair of inputs panned hard left/right or mono (duplicated on both left and right). From what I can tell, that's exactly what I've done. On the Midas, I am either using a stereo track which takes a left and right, nor I'm panning a track hard left, hard right, or center. In other words, I'm not really doing any panning on the Midas.
 
I guess you are saying because I'm panning a track hard left or hard right on the Midas, the pan law would affect it? If so, I wonder why that's not an issue on the commercially available summing mixers? Perhaps because they are just hard wired to left/right?
 
Also, help me understand the idea of leaving the ITB faders at unity and using clip gain for volume; I'm not sure why that matters. 
 
Here's how I calibrated:
 
1. I started with a stereo channel and fed it a stereo signal of the 1k sine wave.
2. The meter leaving the DAW was peaking at -3db with the meter calibrated to -18.
3. The clip light of the channel was lit. I backed down the line input trim until the clip light turned off.
5. I set the Master Output by eye to be on unity (0).
6. I used the Fader of the stereo channel to get the meter reading -3 when the signal returned to the DAW.
 
At that point, I had one stereo channel and the master output calibrated.
 
7. Leaving the Master Output alone going forward, I calibrated the other three stereo inputs the same way.
8. The rest of the channels (save the bass) were sent out as stereo pairs and brought into the Midas on two channels. The left channel is panned hard left, the right channel is panned hard right.
 
I want this to be all as pure of a test as I can make it, but at the end of the day this is about sound, workflow, and fun. If the Midas gives a more open sound than panning in the box, that is an improvement today even if I could go change pan laws in the DAW and try to mimic it. :)
 
 

The stereo tracks would also work.  I am not saying it can only be done with the mono tracks.  What I was really trying to get across was the concept of doing two identical mixes on two different environments. 
 
You could in fact send out 8 stereo pairs for example to the analog mixer that is OK for sure. But all the faders maybe should be at unity in the analog mixer and you adjust the mix by adjusting the clip gains of the audio files perhaps on the stereo tracks themselves.  (Can you send out the track outputs to your Midas after the faders in Sonar) That might be another way.
 
When people are sending out tracks to a summing buss they are doing the mix there.  Either by adjusting the summing buss levels or leaving them all the same and adjusting levels within the DAW from the tracks. That is all good.  But there is nothing being compared to though.  So to say that a mix sounds better just because you are sending tracks to a summing buss is not necessarily correct.  Because you are not doing exactly the same thing ITB  the box and comparing it. 
 
But we are trying to do a test here that is comparing the two things so there must be exactness in terms of the mix levels in both systems for the test to be really valid. This is not happening so far in your test.  I see that there is not a perfect match between those two things mix wise so in a way it is too hard to tell.  The waveforms are just too different.  I noticed that straight away.
 
We need another approach to accurately create the mix on both systems.  The closer they get I am wondering how different they will actually be.
 
Back to my link.  Post 2. All the guys in the room picked the best mix as the one that was done ITB. That says it all don't you think. It does not matter if its ITB or outside the box either.  It still all comes down to how well someone mixes in the end. 
 
And even if consoles are doing things like adding crosstalk etc what about the fact we are into an era now where they are emulating that already. 
 
I used to mix OTB. I had 24 outputs feeding a large Tascam mixer.  I have drifted the other way now and gone ITB. (As also many top name engineers have as well) I feel I am doing my best mixes ever right now. It is not due to them being done ITB either.  It is because I have become a better mixer.
 
You may be wondering why they still do things on large format consoles. It is for other reasons. The Neve summing box has transformers in it. That is what you are hearing. A Neve console has got a transformer on every input channel. That is what you are hearing. The EQ's and things in these large consoles are often well designed and sound excellent. I am not convinced it is the summing part of the mixer that makes them so great. They are now emulating all that with Harrison M32C with that lovely EQ (and UAD) and the SSL channel strips you can get now.
 
In fact I was reading great article in SOS a few months ago in the Studio One column on the CTC-1 console shaper. They tested a whole lot of large format mixers, just the summing parts only. Neve, SSL, API etc. They said there was no noticeable difference between any of them and in fact they all sounded amazingly neutral. That is interesting. That pushed them to look elsewhere in the mixer sound.
 
Too much emphasis in my opinion placed on just the summing buss of any analog device. The sound of the mixer is everywhere else. e.g. the lovely EQ in your Midas etc.. Which we are not using in your test. That is why if you can get two perfectly matched mixes I bet it will be very hard to tell them apart.
 
Don't let me deter you though. Keep on it. It is fun and interesting. I appreciate the effort you are putting in.

Happy Monday Jeff! :)
 
I went back in and used VUMT to calibrate the system. The Dorrough meters just aren't precise enough. I also found and that center detents on the stereo channels of the board aren't always perfectly balanced; on a few I had to tweak the pan pot just a tiny bit to level them out. The mono channels have to have their faders at about +5 over unity to get the signal level proper when using Direct In. After I create another render, I'm going to level match at unity on the pres and try to use the Line Ins to see of the color of the pres adds something or takes away. I did a quick test last night of putting the EQs in all paths, centered (so no EQ in essence) but I thought the mix sounded a bit choked. I will try again with new levels.
 
Yes, I find this incredible fun and interesting. :)
 
Back to my link.  Post 2. All the guys in the room picked the best mix as the one that was done ITB. That says it all don't you think. It does not matter if its ITB or outside the box either.  It still all comes down to how well someone mixes in the end. 
 
I agree - there is no "better" here. However, two different versions of an 1176 comp can both yield slightly different results. Neither is better, but one might be preferred over another by different people. To me, I'm interested in whether or not I prefer to go through the Midas - recognizing that it isn't a magic gooderizer. :)
 
And even if consoles are doing things like adding crosstalk etc what about the fact we are into an era now where they are emulating that already. 
 
I don't think that's the case for cross talk, and I think that may be one of the areas where analog summing can provide something you can't get in the box. For example, Waves NLS does not model cross talk. Slate says they do, but even so this would be simply stereo cross talk - not cross talk from other channels. From what I can understand, in a mixer you can get cross talk from other channels, not just a stereo pair. If that's true, a stereo plugin will not and cannon model that type of cross talk.
 
Here's what I can find on the Midas Crosstalk parameters (not to learn what they mean!):
 
Crosstalk at 1kHz

• Channel to Channel < - 80dB
• Mix to Mix < - 80dB
• Channel to Mix < - 80dB
• Fader Attenuation > 100dB
• Switch Rejection > 100dB

I'm not saying that crosstalk is even a good thing - I'm just saying I don't think it's being fully modeled in the current state of console emulators. Then again, if a console emulator is a good thing for a mix, then why can't running through a console not be a good thing?
 
A Neve console has got a transformer on every input channel. That is what you are hearing. 
 
I am tempted to put the new UAD SLL on all 8 of my outboard stereo buses that go to the board, just to get some of that transformer flavor... Will try it!
 
Don't let me deter you though. Keep on it. It is fun and interesting. I appreciate the effort you are putting in.
 
No worries - I'll be posting more results soon! :)

OK, so I took the time to get VERY precise with leveling all of the channels. I also addressed two issues. The first was to set the bass up in analog summing just like it was ITB. The second was there was a synth that wasn't directed to a mixing channel in the analog summing experiment.
 
You will have to time-align these files to compare - the latency of going outside the box and then back in affects the timing of the file.
 
In the box: https://www.dropbox.com/s/ie57qcr1nwufyfk/ITB%2002.wav?dl=0
Analog summing: https://www.dropbox.com/s/hdrg2ycrjjjownd/Analog.wav?dl=0
 
Conclusion: the differences are so minor that I wonder which files I'm listening to! This method is not worth the effort.
 
Listen for yourself and let me know what you think! :)
 
Next I'm going to try going to Line In and passing through the Midas Pres to see if that adds anything special...

These mixes are much better matched so well done!  Even though the waveforms look much more similar they still will not even null at all which in itself is quite interesting.  But not a big deal. I can time align them very well too. 
 
Yes, like this, the differences are very very small if any.  I am finding it very hard to tell which is which.  That goes to show just using the summing engine of an outside mixer is not worth the trouble.
 
But now that you are talking about Mic Pres sound and of course the EQ in the Midas that will make a big difference I guess.  Mic Pres tend to impart more of their sound as their gain goes up so if your Mic pre are well anti clockwise they may not add much (could end up similar to the Insert INS you were using) but then again they may as well.  If all your Mic Pres had the transformer options fitted things would be very different. But you don't have them fitted I am assuming.  That could be a costly exercise too. 
 
I am right in thinking the closer the mixes the harder it is to tell. 
 
 

Besides the monitoring functions, the summing functions in something like the Dangerous D box could then be considered sales hype?

mixmkr
Besides the monitoring functions, the summing functions in something like the Dangerous D box could then be considered sales hype?

It depends a little on what is inside. If it is just pure resistive summing with a very clean op stage gain stage following then yes maybe it could be all hype. But they don't all do that. The Neve summing box has I think pure resistive summing followed by transformer based gain stages.  I have heard even if you feed a stereo track to the Neve box and back then it sounds different and that means the transformers are adding in the sound.
 
Some of the newer Dangerous stuff has all this colour options built in meaning they are colouring the stereo mix before sending back to you. So yes it will sound different.
 
Fat Bustard II for example has got valves and transformers in the output stages so of course the stereo mix is going to sound different. Why not just sum digitally inside your DAW and send the stereo mix out to some sort of high end valve preamp with transformers in it. Of course it will sound different.
 
A while ago you may remember a Sonar guy built a resistive summing box followed by a high end Mic preamp but it had transformers in it. He put up the two files. I actually preferred the ITB version because it sounded cleaner with less distortion and had better transients going on.
 
 

I have a Stam Audio 1073MPA on order and was considering running through that from the analog summing. But, after coming so close in the test there is really no reason to sum outside of the box. Whether or not running through the Neve 1073 (or a the SSL 4000 I have on order) adds to the process or not - so far there is no need to sum outside the box.
 
I'm going to try summing through the Midas pres tonight and hope to get files up shortly.

OK, new files:
 
This time I went through the midas press - still everything calibrated at unity gain.
 
1. Original ITB file:
https://www.dropbox.com/s/ie57qcr1nwufyfk/ITB%2002.wav?dl=0
 
2. Analog summing through the Midas pres
https://www.dropbox.com/s/e5ydb9t2r1inq6n/Analog%20through%20pres%20unity.wav?dl=0
 
Let me know what you think...