Popslut's scientific "analogue summing vs. ITB test.

[popslut]

I've been wondering about these analogue summing boxes you can buy and whether they actually do anything useful or not.

I've done my research and I've discovered that a lot of far more experienced engineers than me think they significantly improve the sound of ITB mixes. Trouble is, a lot of other far more experienced engineers than me think they're bullshit.

So, no answers there.

The only solution is to hear it with my own ears so, this morning, in between sessions of frenzied masturbation and subsequent self-loathing, I decided to test the theory.

The Test.

I've just spent two days mixing a track for a client in Nuendo. I'm happy with it. The track is split into six stereo subgroups; drums, bass, guitars, synths, vocals and fx and then bounced down to a 44.1 khz / 32 bit float stereo wav.

I've sent each subgroup out to my desk - a Soundtracs Solo Logic 32 - via two RME Fireface 800s - to six stereo pairs made up of twelve identical mono input channels.

Care was taken to calibrate the input gain and fader positions to acheive unity gain. Zero dB at the output = zero dB at the main buss faders of the mixer. This was done with the RME Digicheck application applied to the DAW input channel and apparently accurate to 0.5dB.

The main buss of the mixer feeds a Focusrite Mixmaster Platinum which outputs wordclocked SPdif to the digi-in on the master Fireface, and thence back into the DAW.

I then printed the mix summed through the mix buss of the mixer and compared the two. The levels were very well matched.

I checked the results on Yamaha NS-10s with a Bryston 4b amp, KRK V8s, a 50 quid Pioneer boom box and a pair of AKG K-240 headphones, and just to be sure I was really hearing any difference I asked my wife to switch between channels from the next room using a qwerty keyboard with a long extension so I couldn't hear the key click.

Guess what?

The Result.


I couldn't even tell when whe'd swapped between the versions, let alone hear any difference. They might as well have been the same file.

Just to recap for anyone who missed that particular bombshell - no appreciable difference whatsoever.

Nothing.

Nada.

Not a sausage.

Bugger all.

Not "warmer" or "punchier" or "deeper" or "better stereo imaging" or any of that horseshit. Just the same fucking mix.




A few caveats:

1. I don't by any stretch consider myself to be Mr Golden Ears. I don't pretend I can hear a mouse fart from two rooms away.

2. My mixer wasn't designed by Rupert Neve or George Massenburg. It is, however, well maintained and sonically very adequate. I've done a lot of mixes on it over the last 15 years and I know it inside out.

I did phase invert the two files, just in case there had been a mistake and I was listening to two identical files, and they didn't completely cancel or phase. The residue was mainly top end - hat and snr transients - and could possibly have been due to a slight sample shift caused by the OTB mix passing through 30 feet of balanced cabling, a mixing desk and two stages of D/A - A/D.


Conclusion.

I admit I was expecting at least some difference. Fatter drums or better transient response or something. The only result I wasn't expecting was "no appreciable difference".

At this point, regarding analogue summing, I call "Bullshit".

Now I'm going to mix the track again on my analogue console with all my "vintage" outboard [don't get excited, it's all garbage. Interesting garbage, nonetheless...] and I'm almost certain it'll sound better than the ITB version by a huge margin.

I'll let you know.

[thefool]

this was actually an interesting little read. Guess you just saved me a bunch of money. Thanks for spending your time on testing and writing the results.

Waiting for the next update on the outboard gear!

[thefool]

a side note: i noticed you did the mixing in nuendo. Another interesting question might be (a possible answer was proposed by morerecords once) if summing/mixing in LIVE would make a different result.

morerecords proposed that it sounded like some soft limiting as far as i remember.

would be interesting


edit:on the other hand i think you shouldn't spend time on something i can do myself

[forge]

I did phase invert the two files, just in case there had been a mistake and I was listening to two identical files, and they didn't completely cancel or phase. The residue was mainly top end - hat and snr transients - and could possibly have been due to a slight sample shift caused by the OTB mix passing through 30 feet of balanced cabling, a mixing desk and two stages of D/A - A/D.
.

is there any way you could work out whether that does introduce any latency?

I'd be interested to know whether it cancelled if you found that out and compensated for it

[popslut]

I did phase invert the two files, just in case there had been a mistake and I was listening to two identical files, and they didn't completely cancel or phase. The residue was mainly top end - hat and snr transients - and could possibly have been due to a slight sample shift caused by the OTB mix passing through 30 feet of balanced cabling, a mixing desk and two stages of D/A - A/D.
.

is there any way you could work out whether that does introduce any latency?

It has to, however slight.

I'd be interested to know whether it cancelled if you found that out and compensated for it

I tried nudging it back and forth and it never cancelled either way. There was quite a bit of residue, just not enough to make me want to spend a few grand on a summing box.

To a mathematician or a physicist there would be a difference. I work to pop-music tolerances so, for me, the differences are negligible.

[Precision]

Good read, thanks for posting

[Khazul]

There are alot of very experience analog audio mix engineers around who know fuck all about digital signal processing and its benefits and limitations.

There are only two reasons in my mind to sum in an external analog device:

1. It colours the sound in some pleasing way (odd as they aint supposed unless driven hot).
2. You have within your mix some samples that were bring wall limited an you need to bring the out into the analog domain to (psuedo) recover more of the original signal from the analog overshoots at the clipping points.

Both could probably be dealt with by right plugin if either exists.

2 just happens to be a very cheap and very simple way of dealing with brick wall limited material so it doesnt end up sounding utterly shite after your own limiting, conversion to mp3 etc.

If you get your gain staging right through the digital domain, then TBH - I can see no benefit unless there is something really wrong with your DAW's processing.


OTOH, I can see some very good reasons for using a complete analog mix console rather than just a summing box, all based upon someone knowing that specific console very well, liking its sound/eq character etc etc. this is the reason I keep thinking of going back to an analog console - I just simply miss the damn thing.

[popslut]

OTOH, I can see some very good reasons for using a complete analog mix console rather than just a summing box, all based upon someone knowing that specific console very well, liking its sound/eq character etc etc. this is the reason I keep thinking of going back to an analog console - I just simply miss the damn thing.

Agree entirely.

[substance_g]

The Result.
Not "warmer" or "punchier" or "deeper" or "better stereo imaging" or any of that horseshit. Just the same fucking mix.

Heh, I got exactly the same result when blind-testing a 100 quid mains cable that my hi-fi dealer was trying to convince me was fantastic, compared with the free cable that comes with the CD player. I told him my test results and he said, ah well, of course, you have to listen over the course of some days, not just flick between the two, to hear the real differences. Idiot. He's no longer my hi-fi dealer.

Interesting test and read - thanks for that!

[Sales Dude McBoob]

Pro Tools is the app with the notoriously poor summing engine. Perhaps it was Nuendo that screwed up your test?

[Side note on Nuendo - In my five years working in the MI industry - I have never witnessed a sale, nor met a salesperson who has sold a copy of Nuendo.]

I have my doubts that a summing box would really improve my work. I commend you for experimenting with trying to get more of an out of the box sound. The STUCK INSIDE A BOX thing is really endemic in the Link Your Music section.

[Angstrom]

Thanks for that Popslut - it confirmed my prejudices, which always gives me a warm and fuzzy feeling.

Every time the subject comes up, my hind-brain screams "Buuulllshiiittttt!11".
Fortunately for me - my signal chain and room is so crappy that theres no chance of needing placebo acoustics any time soon.

BTW - I've heard Popslut's mixes and they are annoyingly good.

[forge]

Every time the subject comes up, my hind-brain screams "Buuulllshiiittttt!11".
Fortunately for me - my signal chain and room is so crappy that theres no chance of needing placebo acoustics any time soon.

+1 - was going to say the same - the only monitors I have ar NS-10s driven by a shitty old amp so the mere concept of this kind of test is well out of my reality at present, but it's interestng nonetheless - but I do at least have tissue paper over the tweeters!

the whole debate of sound quality these days just seems like a total fucking waste of time

same as computer power - my mate is out to buy a new iMac and he was getting all uppity over whether he want's a TB of HD space and 2.8GHz core 2 duo or just get the 2.4 and I was saying to him that I seriously cannot think of a single conceivable thing that he might want to do with it that will require that much processor power and he already has several very big external hard drives

totally spoilt for choice

[popslut]

the only monitors I have ar NS-10s driven by a shitty old amp so the mere concept of this kind of test is well out of my reality at present...

As usual, all that audiophile nonsense maens nothing in the real world of popular music.

Next time you find yourself worrying about THD figures or a bit of noise floor intrusion, have a good loud listen to "Pretty Vacant" or "Holidays in the Sun" and realise it's not lack of aliasing noise which gets people's blood pumping.

[Winston]

Every time the subject comes up, my hind-brain screams "Buuulllshiiittttt!11".
Fortunately for me - my signal chain and room is so crappy that theres no chance of needing placebo acoustics any time soon.

+1 - was going to say the same - the only monitors I have ar NS-10s driven by a shitty old amp so the mere concept of this kind of test is well out of my reality at present, but it's interestng nonetheless - but I do at least have tissue paper over the tweeters!

the whole debate of sound quality these days just seems like a total fucking waste of time

same as computer power - my mate is out to buy a new iMac and he was getting all uppity over whether he want's a TB of HD space and 2.8GHz core 2 duo or just get the 2.4 and I was saying to him that I seriously cannot think of a single conceivable thing that he might want to do with it that will require that much processor power and he already has several very big external hard drives

Agreed on the sound quality point. A complete waste of time to worry about these days. Although I disagree about the processing power. I am fine for hard drive space now that I have a 320 GB internal on my MBP, but my 2.16 C2D is still not quite enough for me to feel like I can throw anything on any track that I want. I like to keep things non-bounced and I like processor hungry instruments and effects. Once a quad comes out, I'll be very comfy.

And also agreed on the hardware mixing vs In-the-box. The only thing I'm willing to admit hardware will provide over software is analog overdrive. I only know this from being a guitar player. When analog gear is driven to overdrive it can give a sound more pleasing to many than software counterparts. But that difference is narrowing too. I can imagine that some desks out there are praised for that overdrive characteristic, but I'm still more than happy to throw my favorite saturation/overdrive plugs on a track for that .

Nice test btw, thanks. Now would you mind testing Live 7's 64 bit marketing...er. summing against Live 7's measly 32 bit quality?

[Angstrom]

oh indeed, there are some things that analogue can do that digital still can't get right. Not that it's impossible, but that it hasn't been done yet.

I have a recording made using a B&O tape delay that just sounds berserk ! There's no way I could recreate it in software today.

but in the context of adding one signal to another without modifying the gain, my predjudice is that digital and (high quality) analogue are on equal footing. When we start to add colouration - that is a different story.