STM2448 and SX Mixer Comparison Test and Results!!!

rodos1979 · Post by **rodos1979** » Thu Feb 13, 2003 6:54 am

Hello to all!

There were some recent discussions about mixing in SX and mixing with the CW mixers (STM2448 & STM4896)and which of the two is more accurate.

I went on and I did a test. I am going first to explain you what was the test about and then I am going to give you the results.

First I chose a commercial CD Track and I copied it to my harddisk. The track I chose is called "Pompton Tunpike", is performed by the BBC Big Band and is track 3 in the CD: "The Age of Swing Volume 4" (Tring International Plc, VAR100) As you can see it is a big band piece and it has a lot of high pitched brass instruments and a clarinet & muted trumpet solo. The reason I chose such a track is because it is more easy to trace phase problems on recordings of brass instruments with sharp attacks (as Astroman had suggested somewhere else in the forum).

Then I opened a Cubase project, loaded the piece to 8 different stereo tracks, all panned at the center and at unity gain. I had to reduce the Master Mix fader by -19db in order for clipping not to occure. I recorded the output of the SX mixer to a stereo CD quality track and saved the file.

Then I loaded the STM2448 mixer and routed the 8 different cubase tracks, each one to a different channel at the STM mixer and disabled the rest of the channels not in use. All the tracks were with the same panning/volume settings as in the Cubase Mixer. I reduced the Master mix fader by -19db as I had done when mixing with the SX mixer and recorded the output (MixL&R) of the STM2448 mixer to a CD quality wave file.

After that I did another recording as the previous one but with the phase compensation set to ON. Therefore, in the end I had 4 files:
a) the original cd track
b) the sx mixer track
c) the stm2448 mixer track, phase.comp off
d) the stm2448 mixer track, phase.comp on

I opened all files in Wavelab 4 and run a Global analysis for each one of them. And copied the results to a Word document. I realized that the -19db attenuation was just a bit more than what was needed so, I normalized all tracks to peak at 0db. Then all files, sounded *almost* the same in volume. I did again a Global analysis of all the files and saved the results to the same Word document. All results were organized to a table so that the various values can be easily compared. Plz download the table so you can check the various values I am refering at. (all files referred to can be found at the web address in the end of my thread)

I am going to refer only to the normalized files as these are more "comparable". First have a look at the various values of the Original CD Track. Now, have a look at the STM2448 Normalized with phase.comp OFF values. They are a bit different aren't they? So, it seems that this track is not a real perfect copy of the original CD Track. But, believe me, there is no need to compare numbers in this case. A quick listen to this tracks reveals that it SUFFERS from PHASE PROBLEMS! It feels slightly lower in volume and there is a considerable loss in high frequencies. Furthermore, it sounds a bit muddy.

Now, have a look at the STM2448 normalized with phase.comp ON values. The values seem almost the same. The differences are so small that are negligible. And oh yes, it sounds the same to the original (at least through my mid-quality *hi-fi* speakers). However, since there is a slight difference in the values, one cannot say that this is a perfect copy of the original.

And now have a look at the SX mix normalized track values. Amazed? The values are EXACTLY the same! A perfect copy of the original CD track! Oh yess, it sounds *exactly* the same too (again through my speakers).

Finally, I played the 4 tracks through Waves PAZ Analyzer and saved the results to 4 different files you can download and see a comparison of the peak values of all the frequencies in each track.

CONCLUSIONS:
1) The STM2448 with the phase compensation button set to Off SUCKS!
2) The STM2448 with phase compensation On is almost identical sounding to the Cubase SX mixer, with the latter being the winner with a very slight difference.
Of course, these things apply when mixing 8 stereo tracks. I dont know if the same mixers behave in the same way when mixing 24 stereo tracks or more.

I hope this has helped and cleared things a bit.

Thank you

P.S. "Have you set the phase compensation button to ON, my dear???"

P.S.2. I am sorry but at the moment the Geocities Page Builder is not working, so I cannot upload the files. I ll do so the soonest possible!

[ This Message was edited by: rodos1979 on 2003-02-13 07:02 ]

Post by **hubird** » Thu Feb 13, 2003 11:23 am

Hi Rodos, it looks like you did a great job.
I never gave special attention to the phase issue, but if I read your report of the test I should.
Said that I checked the manual and looked in channel strip of the STM 24/48 mixer for the phase compensation 'switch'.
It's called 'delay' and has a range from zero to 200. That's the one?

So if I'm right I have to experimentally search for the right number of samples to delay per channel, while listening to the result?
Should I do this for all the channels, or does it depends on the casual material in a channel?
Well it won't be easy to be sure not to make it worse than it was.

Another complicating thing is that, concerning the phase compensation, the channels seem to be grouped per three: 1-2-3, 4-5-6-, etc., as tthe manual explains.
What to do with this?
And why are they combined per three and not per 4?
I suppose this way combined stereo channel handling of a splitted mother channel gets possible.
But how to deal exactly with it in practice?

Don't blame the messenger indeed

but there are so many questions beyond the pure level of the test, I'm wondering if it's wise to play around with it without really knowing what to do.

Besides, without any doubts about the quality and the technical results of your carefully explored test, when I make a bounce from one or more Cubase track(s) via the STM 24/96 mixer back to Cubase, I've never heard any difference between the old and the new track (I always compare them carefully to bring the volume level of the new track to the old stage).

So my conclusion is this:
I'm mutch too far away from knowing how to handle phase compensation.
The test can't learn me that.
So I hope here on planetz are some folks who have the knowledge to tell us how to deal with these issues in practice.
Coz that's what it's all about, isn't it?
cheerz.

_________________
Let There Be Music!

[ This Message was edited by: hubird on 2003-02-13 11:25 ]

King of Snake · Post by **King of Snake** » Thu Feb 13, 2003 12:04 pm

Yeah don't be too quick to base decisions and conclusions on one person's test. I've seen so many test been done concerning Cubase/Nuendo summing busses and mixing, and each had different results and yielded different conclusions. The best thing to do (I guess) is use your ears. Then again, if you're not very experienced (like me) you migth not trust your ears all the time, especially with crappy speakers. (I'm thinking of getting a pair of Mackie HR624's tomorrow. Should make a lot of difference for my mixes

Although I really can't afford it, I may just have to buy them anyway, I'm really sick of trying to mix on cheapo hifi speakers)

kensuguro · Post by **kensuguro** » Thu Feb 13, 2003 12:14 pm

I've noticed the phase issue.. I've had to work around it in some of my mixes, but my question is, how much phase shift has occured? It it some random number depending on the way the DSP is allocated? Or is it always the same amount of shift occuring on the same tracks?

visilia · Post by **visilia** » Thu Feb 13, 2003 12:21 pm

Hi Rhodos,

I think you did a great job with the test. I really would like to see/hear your results. But your explenation allready makes perfectly clear that I really have to be carefull when I want to setup a test about the subjective audio quality of the CubaseSX and SFP mix engines.

However, I think that these phase issues hardly will be a problem in practice, because you normally wouldn't have several channels containing the same audio signal. A problematic situation I can think of is when tracking an acoustic source with multiple microphones. Say you're recording a drum kit with one mic for the kick, one for the snare and a stereo overhead for the cymbals and toms. There's probably a lot of crosstalk on the overheads from the snare in this situation. So if you put everything together in the STM2448, the phase difference could have an impact on the sound of the snare. At the other hand, the distance between sound source and microphone can introduce these exact same issues, so I'm not sure if the STM2448 phase issue will be a problem.

What really would be bad is if there would be phase difference inside stereo channels of the STM2448. I'm still not sure if this can happen.

Hubird, the phase compensation switch isn't the delay parameter of channels. It's one button somewhere on the master panel of the STM2448 mixer that should make sure that there aren't any phase differences between channels, but Rhodos just made clear that this function doesn't work 100%.

cheers,
vincent

Shayne White · Post by **Shayne White** » Thu Feb 13, 2003 12:27 pm

It's really simple to activate phase compensation. There's a "PhaseComp" switch right next to the master fader if you have Options selected. Enable it, and that's all you need to do. That "Delay" slider is only for adjusting individual instruments that are out of phase with each other from the beginning (such as stereo micing a guitar). Most of the time it's not needed.

By the way, you only need to activate PhaseComp if you have a single sound source going to multiple channel inputs. If every channel has a different sound source, then there is no need to activate PhaseComp since every instrument is different to begin with.

Shayne

Grok · Post by **Grok** » Thu Feb 13, 2003 12:38 pm

The phase compensation button is useful when doing multimics multitrack recording in one time in a given space...

1st example: recording a real drummer

2nd example: recording an orchestra with two stereo microphones

3rd example: recording an orchestra with a quadraphonic mic like the Soundfield mic

The phase compensation isn't useful when doing stereo or mono overdubs. It isn't useful when recording a complete MIDI sequence with stereo or mono synths and samplers.

_________________
Toujours l'Amour!

[ This Message was edited by: Grok on 2003-02-13 12:47 ]

Grok · Post by **Grok** » Thu Feb 13, 2003 12:43 pm

On 2003-02-13 06:54, rodos1979 wrote:
...
CONCLUSIONS:
1) The STM2448 with the phase compensation button set to Off SUCKS!
...

Nonsense

_________________
Toujours l'Amour!

[ This Message was edited by: Grok on 2003-02-13 12:45 ]

Grok · Post by **Grok** » Thu Feb 13, 2003 1:02 pm

On 2003-02-13 12:14, kensuguro wrote:
I've noticed the phase issue.. I've had to work around it in some of my mixes, but my question is, how much phase shift has occured? It it some random number depending on the way the DSP is allocated? Or is it always the same amount of shift occuring on the same tracks?

I think the answer is in your question

If it was "always the same amount of shift occuring on the same tracks", then CW developpers had surely corrected this "same amount"...

So I guess "it is some (not random) number depending on the way the DSP is allocated"...(if it was random numbers, then no phase compensation would be possible)

Cheers,
Grok

_________________
Toujours l'Amour!

[ This Message was edited by: Grok on 2003-02-13 13:03 ]

visilia · Post by **visilia** » Thu Feb 13, 2003 1:03 pm

On 2003-02-13 12:43, Grok wrote:
On 2003-02-13 06:54, rodos1979 wrote:
...
CONCLUSIONS:
1) The STM2448 with the phase compensation button set to Off SUCKS!
...
Nonsense

_________________
Toujours l'Amour!

Not if you see it in perspective. I think you can conclude that the phase handling of the STM2448 does suck, but as I tried to explain in my previous post, this won't be a problem in practice most of the time, so the STM2448 does NOT suck....errr.... I'll shut up now

Grok · Post by **Grok** » Thu Feb 13, 2003 1:05 pm

Post by **hubird** » Thu Feb 13, 2003 4:08 pm

thanks everybody, it's more clear now for me.
At least I found the damn button (shame on me, but I was on the wrong track

).
Hmm, one sound over more channels...two mics catching one sound...time delay...phasing...I got it.

He he, I have just one mic, and I don't really like natural drums...so... lucky me

kimgr · Post by **kimgr** » Thu Feb 13, 2003 4:46 pm

It's a great idea to do a test of phase coherency on the SX and STM mixers, BUT:

Like SOOO many tests before this one, it tells you absolutely nothing about the sound of the mixer... This test has nothing to do with a real-life working situation.

Why would anyone care what a given mixer sounds like with all the faders set at unity gain, and all pans set dead center or hard left/right only ??? And without any plugins or FX's ? And playing the same audio on multiple tracks ?

The idea of having a digital mixer that's mathematically correct is absurd, except for some very special mastering applications.
It's not even possible to create such a mixer, and even if it were it wouldn't (subjectively) sound good. In fact, I wouldn't "sound" at all.
Just like in the analog world, every time you alter the original signal in ANY way, you actually "distort" it. The question is just how, and how much you distort it.

Unfortunately no one knows exactly how the different audio engines/mixers handles the signal throughout the signal chain, but there is one fundamental difference between the SX mixer and the STM mixers: The SX mixer is using floating point mathematics, and the STM's are using fixed point.
To my ears, fixed-point @32bit or higher sounds better than any "floating" mixer I've come across.

I've done extensive tests of different digital mixers in the past, but instead of looking at mathematical correctness, I focused on listening to the end results.
And I used "real-life" songs with 24 to 48 channels, and mixed them with plugins and FX's.
Most importantly, I used blind testing when evaluating the results.

Here's a list with my current rating of the mixers I've tested:

1) Pro Tools|HD dithered mixer
2) Creamware STM mixers
3) Ensoniq Paris
4) Emagic Logic Audio
5) Mackie d8b
6) Pro Tools LE
7) Pro Tools|24 mix

Cubase/Nuendo

Happy Mixing,

Kim.

rodos1979 · Post by **rodos1979** » Thu Feb 13, 2003 5:06 pm

Hello to all!

Ok, there are some things I should make clear in order for misunderstandings not to happen. (I do not want any fanatic STM2448 users to burn my car!)

First of all, I want to assure you that I did the best I could, to my knowledge, so that this test is as accurate and correct as possible. I hope nobody questions that

Second, this test is certainly not a REAL LIFE scenario. Who in the world would have the same track multiplied by 8 times and why? It is just more simple to push the fader up or turn the gain to a higher volume. As in theory and in practice, multiplying the exact track by 2 or more times (8 times in this case) leads to a higher level track (that is what Physics say about adding waves that are *exactly* the same).
HOWEVER, my test clearly showed that what should have happened theoritically, DID NOT happen in practice. The STM2448 with the phase compensation Off produced a track of lesser quality due to phase cancellations, especially in the range of 1500Hz to 8000Hz.
Third, Visilia stated above that I showed that the phase compensation of the STM doesnt work 100%. This is not exactly the case. I have clearly stated that the SX track and the STM (ph. ON)track both sound identical to the Original. It is just that by looking the numbers one sees that the STM values are slightly different (but as I said the difference is SO small that it is negligible). The STM2448 is a wonderful mixer (99.9% as good as the SX mixer, in the context of this test) but should only be used with the phase compensation button set always to ON, in my honest opinion.
Fourth, the real conclusion of this test is that the STM2448 mixer DOES colorize the sound (especially when phase compensation is set to off), while the SX Mixer DOES NOT. This is NOT NECESSARILY a BAD THING! Analog machines are praised for this exact property they have. The question is if the colorization works "FOR" us or "AGAINST" us. In my honest opinion, the STM2448 with the phase compensation button Off can be an "ENEMY" (again possibly not in all cases, as this test IS NOT a real life scenario).
Let your ears to make decisions for you.

All the relevant files can be found at http://www.geocities.com/rodos1979/pulsar.html. Have a look especially at the Word Document, as well as to the Comparison Graph I just added (where things are clear as water). This graph was made using the iZotope Ozone 2 mastering program (a must have!). The graphs represent the average values in db of the whole frequency spectrum of each Track. Note that the blue graph of the Oroginal Track is not visible at all as it resides exaclty under the yellow graph of the SX Track. Again, note that the yellow line is almost not visible at all as it coincides with the purlpe line of the STM ph.comp.ON Track. Have a look in the 1kHz to 15kHz range to see just some yellow pixels emerging now and then away from the purple line. And finally, have a look at the white line of the STM ph.comp.OFF Track. It is clearly visible almost in all the frequency spectrum. Note especially how much it differs in the 1.5kHz to 8kHz range. These frequencies have been cancelled out due to phase problems.

I don't know how useful can be my test. As Hubird said what matters after all is to make better music. I dont have the knowledge to explain the test and draw real musical conclusions of the type of "do that" or "avoid doing the other". Therefore, I would like the more wise and experienced people of this forum to try and draw some "musical" conclusions out of my test and enlighten the rest of us. I have already done what I could do at the time being and with the knowledge I have at the moment.

Thank you

[ This Message was edited by: rodos1979 on 2003-02-13 17:09 ]

otter · Post by **otter** » Thu Feb 13, 2003 5:14 pm

@ KIM:agree.
Judge by the two things with the hole that are attached to your head on both sides.

Close the two things that are slightly above your nose.mind the difference between -video- and -audio-

btw imho the STM mixers sound way better than the SX mixer.
@rodos1979: respect for your good work.

still is interesting,anyway !!!

[ This Message was edited by: otter on 2003-02-13 17:15 ]

[ This Message was edited by: otter on 2003-02-13 17:19 ]

rodos1979 · Post by **rodos1979** » Thu Feb 13, 2003 5:23 pm

Hello again!

Returning to view my edited post I saw that meanwhile Kimgr had posted his view on the subject.
Kim, it is very nice to let us know your personal view on the subject. I respect your knowledge and experience and I agree with you as you can see if you read the post I was editing as you were posting your answer.
If YOU say that the STM mixers are the 2nd best digital mixers you have tried, then I take your word as an axiom.
I just would like you to explain in which other cases the STM mixer could behave in a similar "problematic" way, except in this test's case. Furthermore, what can be done to avoid possible phase problems when mixing with the STM.

Thank you

astroman · Post by **astroman** » Fri Feb 14, 2003 4:49 pm

hi rodos,

imho your efforts are a great hand's on experience about some aspects of digital mixing.
My 2 cents about what's happened:
actually the SX hasn't processed anything, like Kim already pointed out - and unfortunately that magic number '8' was rather contraproductive. You know it's one of the numbers that computers handle really well, so dividing by 8 in native CPU math always yields great results.
SX certainly recognized that no processing was due, so switched to an optimized path without the 'sound processing engine'.
In SFP the setup of 8 same channels (might have) led to distribution over several DSPs with that ultra-small sample delay which you heard as frequency degradation.
The phase switch seems to compensate that according to your results.
I suggested to use the same information on multiple tracks (2 would have been enough btw) as a kind of indicator because even small inaccuracies get hearable immediately.
But those setups are difficult anyway: if you apply an eq gain in stage 1 and later the opposite in stage 2 you won't get back the original because the signal level and spectrum have changed in between.
One important notice: you used a CD track, which means you can't get beyond that already digitzed quality, whatever it was.
I used to have a copy of Billie Holiday's Lady in Satin (CBS - digitally remastered from the original analog tapes - as the label proudly states).
Recently I bought the audiophile vinyl version which is way superior even on a Dual 701 with Ortofon OM20E and a Pro-Ject Phonobox preamp - I'm no high end fanatic

It would be better to generate some high quality sounds in SFP or do some acoustic recordings.
On the other hand while I trust in Kim's chart because he's experienced and works on a professional level, it also helps to 'sharpen' your ears if you do some tests yourself.

cheers, Tom

six_wax · Post by **six_wax** » Fri Feb 21, 2003 7:35 pm

you kids may find this to be of interest:

http://www.musicgearnetwork.com/ubb/ult ... 004754;p=1

rodos1979 · Post by **rodos1979** » Fri Feb 21, 2003 8:55 pm

Hello!

It is too late and I am too sleepy to understand what are they talking in that forum. However, they seem to know a lot. Let's wait for the results of their test!

Thank you