Passive Summing Box -- Blind Listening Test

I built a passive summing box. Based on the New York Dave design.  24 inputs.  4 mono channels and 10 stereo pairs.  Hard wired, no switches.  10kR 1/2 watt resistors.  243R 1/4 watt output resistors.  

I'm trying figure out what I think the results actually are and have concocted a blind listening test to see what people think without their prejudices getting in the way. What I've done is I took a track I mixed recently for some friends of mine and made several new mixes. I was particularly interested in isolating the effect of just running the tracks through an extra D->A and A->D conversion with a mic pre in between, versus actual analog summing, so I have one mix that is the ITB mix run through the passive summer and a Sytek preamp in addition to a mix that actually bust instruments out onto separate channels before running through the summer. I also, of course, compared the sounds of different preamps in my collection.

The mixes are:

- ITB, everything mixed in the computer
- Mixed ITB, run 2 channels through Sytek
- Mixed OTB, Sytek
- Mixed OTB, Sytek Burr Brown
- Mixed OTB, RNP
- Mixed OTB, Great River
- Mixed OTB, Hamptone Tube Pre

Since I had previously done this mix entirely ITB and happened to have a software compressor across the drums, I took the lazy route and brought the drums out as a two channel stem rather than mix individual drum mics in the passive box. So on the tracks mixed OTB, what you're really hearing is the drums, bass and guitar mixed as stems. There is a spring reverb on the guitar and a software reverb on the drums that each got their own channels on the summer as well.

I have then taken the first 30 seconds of each mix, shuffled randomly, and posted them here:

http://spacelabstudio.com/passive_summing_fixed/

Anyone who would like to play should just reply to this thread. I'm interested in hearing:

1) Which version to you think sounds the best.

2) Can you describe the differences that you hear between the versions?

3) Which wav files do you think are which mixes? Can you match them up?

It would be great if you didn't read other people's responses before taking the ITB vs OTB challenge.

Thanks all!

Chris

It is with deep embarrassment that I am forced to admit that in the rush to play with my new toy I completely failed to notice that one channel was out of phase with the other in all of the mixes that passed through the summing box. I apologize humbly for having had you listen to f'ed up mixes. Listening this morning, the error was pretty obvious.

I have used a wave editor to flip the phase in the right channel of each wav file. And I have rerandomized the ordering, so they are now in a different order again. I have changed the url above, but here it is again:

http://spacelabstudio.com/passive_summing_fixed/

Sorry about that.

Chris

Last time I tried to do a listening test with ITB vs OTB, a war started. I hope this time turns out to be OK! ;D

The more variables between the two samples the less definitive any blind listening comparison. It seems two mixes made on the same platform will typically sound slightly different. Further, if you are listening to the result of the mix as you make it, you could correct for some apparent path differences in the course of mixing.

Good luck.. 

JR

JohnRoberts said:

The more variables between the two samples the less definitive any blind listening comparison. It seems two mixes made on the same platform will typically sound slightly different. Further, if you are listening to the result of the mix as you make it, you could correct for some apparent path differences in the course of mixing.

Good luck..   

JR

Click to expand...

I'm not touching any knobs when I record each mix.  The mix is done and then it just gets routed through the summing box or not.

Chris

You should edit your first post to delete the wrong link - people might not read the second post if they want to do the blind test!

Hi,
Interesting test...I listened to a few. I would say that #1 doesn't have as much clarity as the others. The difference in very subtle.
Also there aren't that many layered instruments and they already seem to be taking up their own sonic space so that makes the Summing box even more transparent.

Also, summed in or outside the box isn't going to help the fact that the whole song starts to drag a few measures after the drums come in.

EDIT: On second listens the differences are so subtle I think sitting hear trying to split hairs that are soo tiny my brain will probably effect more of what I'm hear than what I'm actually hearing
But a very cool test!
Thanks for the ear candy-ness

dmp said:

You should edit your first post to delete the wrong link - people might not read the second post if they want to do the blind test!

Click to expand...

I did that when I posted my correction.  Thanks!

Chris

I can't tell the difference! I listened to them all twice.
Thanks for posting.

These tests are always good to do. 

However, I can tell you that before I listen to any files I'm already hopelessly confused by your post.  Everything gets way too fuzzy after your first paragraph.  It leaves many questions open, like:

Did you use all the passive mixer channels in each test?  If not, did you put a resistance across unused channels at the XLR-in to prevent level changes? 

Are all levels and plugins exactly the same in each test? 

Are all pans, etc exactly the same in every test?

What are you trying to say about the drums and all the stems??  You lost me.  If you are using premixed ITB stems, then this partially defeats the purpose of your test!  :  Did you run out of passive channels?  How many channels are premixed to stems for each instrument?

What are you trying to say about the spring reverb?  Is this an analog box or a plug in you're talking about?  You lost me again.  If you are changing FX between tests you are partially defeating the purpose of your test!

How did you control final levels between tests?

What ITB software did you use? 

Did you bounce to disk or did you run in real time?  How did you bus your channels?

Are all FX exactly the same ITB and OTB?

Are all tracks automated and run exactly the same ITB and OTB?

Are all tests using the same converters?  How did you control consistency between various converters and preamps?

ETC!

Changing ANY of these variables has the potential to change your results!

Therefore you are no longer hearing a strict comparison between ITB and OTB. 

Your reliance on premixed stems already has the potential to spoil the major benefit of OTB mixing.

One needs to be extremely careful and scientific about these types of tests, otherwise they become anecodotal evidence rather then empirical evidence.

I don't get it either. Why not just two mixes, one summed ITB one through the summing box?

I would also recommend to get the most out of your OTB (if you have enough out channels) split as much stuff to use alll your outputs.

Allll your drums (welll kick and snr you can usually get away with going to one channel since you rarely have them going on at the same time....) but like definitely VERBS and DELAYS and Overheads and all that good stuff....

2 pesos

tommypiper said:

Your reliance on premixed stems already has the potential to spoil the major benefit of OTB mixing.

Click to expand...

correct.
file 4 and 7 sounds nicer than the rest in my opinion.
i would bet file 3 is the mixed ITB one
oh and i listened with an MDR7506 plugged in my grey aluminium computer

Sorry I didn't get back to this sooner.  Of course, I thought I'd be sparing you, gentle reader, by sparing you the gory technical details of the set up, but apparently you have the probably reasonable expectation that I demonstrate I'm not a complete moron.

tommypiper said:

Did you use all the passive mixer channels in each test?  If not, did you put a resistance across unused channels at the XLR-in to prevent level changes? 

Click to expand...

All of the passive mixer channels remained connected to the DACs which were driving them.  The number of DACs with signal coming out of them was different for the test where I ran the ITB through the box, but the electrical connections remained identical, so no level changes from one test to the next.

tommypiper said:

Are all levels and plugins exactly the same in each test? 

Click to expand...

Yes, of course.

tommypiper said:

Are all pans, etc exactly the same in every test?

Click to expand...

Yes, again.

tommypiper said:

What are you trying to say about the drums and all the stems??  You lost me.  If you are using premixed ITB stems, then this partially defeats the purpose of your test!   :  Did you run out of passive channels?  How many channels are premixed to stems for each instrument?

Click to expand...

It was an initial test and I was being lazy.  The drums were already being submixed in the ITB mix with a buss compressor across all of the drums.  So it made sense to take the output of the drum bass as a single stereo track.  I probably should have waited until I had a better test before I bothered anyone in public.  I apologize for that.  See next post for a better test.

tommypiper said:

What are you trying to say about the spring reverb?  Is this an analog box or a plug in you're talking about?  You lost me again.  If you are changing FX between tests you are partially defeating the purpose of your test!

Click to expand...

The spring reverb is an analog box that was for the ITB mix as an external effect, so out through two DAC channels and back in through 2 DAC channels.  To prevent the levels in the OTB mix from being different than the ITB mix, I then just routed the reverb channel back out again via two DAC channels for the OTB mix.  So the reverb was being abused mightily, going D->A->D->A, basically.  I opted to eliminate it altogether for the test in the next post.

tommypiper said:

How did you control final levels between tests?

Click to expand...

I attempted to do a group normalization with Audition.  It didn't work very well, as you can hear in the files.  In the next test I did this in a much more controlled way--by running a test tone through a mono input in the box I adjusted the preamp gain levels to give me end to end unity gain in each channel (within the tolerance of the meters in Reaper).

tommypiper said:

What ITB software did you use? 

Click to expand...

Reaper.  I believe it does 64 bit floating point arithmetic so, in theory, rounding error should should be so far down that you wouldn't notice.

tommypiper said:

Did you bounce to disk or did you run in real time?  How did you bus your channels?

Click to expand...

Recorded output in real time.

tommypiper said:

Are all FX exactly the same ITB and OTB?

Click to expand...

Yes.

tommypiper said:

Are all tracks automated and run exactly the same ITB and OTB?

Click to expand...

Geez, yes, already.  You must really think I'm a dumb ass! 

tommypiper said:

Are all tests using the same converters?  How did you control consistency between various converters and preamps?

Click to expand...

Converters were the same for each test.

tommypiper said:

Changing ANY of these variables has the potential to change your results!

Click to expand...

Duh!

tommypiper said:

Your reliance on premixed stems already has the potential to spoil the major benefit of OTB mixing.

Click to expand...

Yep, see next post.

Chris

So, the first test was probably not fair in that it used a stem for the drums.  I've conducted another test with a different piece of material where for the OTB test I'm sending all of the tracks through individual channels in the summing box.  I have also gotten rid of the idea of testing multiple make up gains.  To see what signal sounds like through different preamps, just run signal through different preamps.    In this test I'd rather focus on just the use of passive summing, controlling everything else to the extent possible.

There are four files in this test, randomly shuffled:

- ITB mix
- ITB mix run through a single stereo input on the summer
- OTB mix, separate tracks all run through the summer.
- One of the above, chosen randomly and copied.  Kind of a spoiler.

The ITB mix and OTB mix utilized the exact same levels, pans and effects.  To obtain the OTB mix I simply assigned individual tracks, post fader, post effects, to either a mono are stereo channel on a DAC running into the passive summer.  Kick, snare and bass all got a single mono input on the summer and floor tom, rack tom, stereo overhead (MS), left guitar, and right guitar each got their own stereo input on the summer.  The output of the mic pre used for make up gain (Sytek, non-Burr Brown channels) was brought back in via an ADC and recorded.

The only postprocessing on these files is I used r8brain to convert from 96khz/24bit to 44.1khz/16bit files, so that you had a hope of downloading them today.  I established even levels by running a test tone through a mono input on the summer and adjusting the preamp gain to give me end to end unity gain on each channel.  (Within the limits of Reaper's metering resolution--about a tenth of a db.)

Let me know if I can provide more information about how the samples were recorded. 

What do you guys hear?

http://spacelabstudio.com/passive_summing2/

Thanks!
Chris

PS Oh, I almost forgot, in case anyone cares, the answers to the first test:

1) OTB, Sytek
2) ITB
3) Sytek w/Burr Brown chips
4) Great River
5) ITB run through Sytek
6) ITB run through Sytek (identical file to 5)
7) OTB, RNP

Thanks.  The new mix descriptions seem a little bit clearer.    You still haven't answered my question, how many channels are you using?    I believe we'd hear a difference between 24 OTB channels vs ITB.  However, 8, 10, 12, may not be so apparent...  Reaper mixing at 64 bit may also sound better than PT and others that use 32 bit.  So that's of interest.  Also, running in real time to mix ITB may improve the fidelity.  So you've got a potentially narrower gap between ITB and OTB.

I'm on a MacBook laptop with no headphones so I won't even try to listen to the files until a later time when I can use good headphones..

good luck.

tommypiper said:

Thanks.  The new mix descriptions seem a little bit clearer.     You still haven't answered my question, how many channels are you using?    

Click to expand...

spacelabstudio said:

Kick, snare and bass all got a single mono input on the summer and floor tom, rack tom, stereo overhead (MS), left guitar, and right guitar each got their own stereo input on the summer. 

Click to expand...

Three mono and five stereo, so 13 total.  Although I have 24 inputs (4 mono, 10 stereo) only 16 of those (4 mono, 6 stereo) are actually connected to converters, so electrically it's a 16 channel unit at the moment.  Just need to solder some more cables into my patch bay before I can use all 24.

FWIW, I had convinced myself I could hear a difference this time,when unblinded.  I just gave myself the blind test, though, thinking I had a chance at picking which mix was which, given that I had already heard them several times unblinded.  Turns out I had no clue.  Go figure.

Chris