Basic stereo setup, how do you do it, what is important to you?

[terryj]

someone posted in a thread somewhere sometime (precise eh!) that on the net was some sound samples. IIRC it went something like 'two sounds, one played at a much lower level but preceding the other louder one and yet the sound seems to come from the speaker playing the lower level tone.'

I only mention it in the hope it will ring someones bell, and we can get the link. Sounds like an interesting demo if it exists.

 

[garylkoh]

My only 'concern' with using tracks and saying 'this is what should happen' is that (not you, but I'd love your reaction to this) *we* get used to 'our sound'. And, being arrogant bunch that we are, what we hear is, naturally, correct. Hence, if another systems does it differently the danger is that it (when it might be right) can be labelled wrong, because it is not how ours does it.

This is why I said that this is music that *I* have used consistently for years and years. I have heard the same music on a sufficient number of systems to be confident enough that that is the correct interpretation of the music. I am also blessed by being a friend of Winston Ma who recorded most of the music in the list and I have heard the music in his room - which is the reference.

Middle C = 261.63 Hz so at 1127 fps the wavelength is 4.3076’ = 51.69”
Assuming we resolve to 0.200”, then 180 deg x 0.2" / 51.69" = 0.696 deg

Actually, 360 degrees per complete cycle, so about 1.4 deg.

There has been a lot of new research since 2001 on how we hear - and one of the theories is that the brain localizes sounds by comparing the phase difference of different formant frequencies of both ears. This trigonometric exercise only focuses on one frequency - middle C. The brain/ear actually uses a combination of frequencies distorted by the folds of the outer ear so that you can localize not only left to right in front of you, but behind you, and height and distance. This is why your image localization goes haywire when you have an injury to your ear that deforms it.

Since you're downloading stuff off my website, I also have a couple of newsletters where I have articles on How and What We Hear - and references to research papers on the physiology and psychology of hearing. Knowing how we hear gives us a much better reference point to stereo speaker set-up than the usual equilateral triangle/golden ratio rules of thumb.

What I try to achieve is that the left and right speakers define the "window into the soundstage". The window is shaped by the speakers, and hence distance, height, camber, tilt, and toe are all important.

 

[terryj]

well, inspired by the thread (who WAS the Op anyway?? he must be a genius haha), and given that my music chair has been taken back by the manufacturer..to be done properly we hope..I could be a lot more accurate in my physical measuring. With a tape measure. On the floor. Without a chair in the way.

It is almost embarrassing to admit how far out I was...but for the sake of illustration and honesty I'll tell you and just cop the laughs that will follow, and fair enough too. (can I even admit that it never occurred to me to lay the tape on the floor as Don told us he did???)

I thought I was pretty close, but the imaging has been driving me mad! And, to my horror, I found the left and right differed by about three inches. (hangs head in the shame of it all).

So, just to backtrack, I used (audio) measurements to hone in on this. I had assumed I was close physically.

I used REW to make a sweep of left and right channel, then looked at it in the time domain (most people, and until rather recently myself as well) only ever look at FR.

And only ever 'fix' things with eq. Try telling them that time based problems cannot be cured by eq...........anyway I digress.

The 'normal long range shot' of the impulses

then by changing the graph limits..should be self explanatory..here it is magnified and close up

Hold crtl and shift, then right button on mouse leads to a direct time and distance readout.

Adjust speakers (or chair) accordingly.

Anyway, what prompted me to start the thread (before I found out how wrong I was initially with the physical setup...oh the shame) was how much that timing difference affected things. Sure, it's obvious that it would (and why gary for example stresses in his how to that you need to be down to one mm accuracy) but until you actually experience it you kinda don't know...if you follow.

that was why I was asking 'theoretically' whether it's a case of 'close enough is good enough' etc etc.
I have the advantage of being able to 'cure' that time delay by setting a delay on the appropriate speaker in the deqx, and time align them that way.

It also meant I could leave that delay in another profile, and switch between the two with a push of a button on the remote.

I gotta tell you, the difference is quite amazing (hence the thread, I was so 'stunned' by it). As an example, the time difference between the two channels was 200/1 000 000 of a second. Spose I should reduce that, but hey I'm not in school now!! Ok, 2/10 000 of a second difference.

The shift in the image was easily eight inches or so, and to bring it back to centre using vol offset required around 2 db! That seemed staggering to me....until I read gary's how to which goes into it.

And of course the difference did not stop there, the centre image felt far more natural (done time wise-ie properly-than with blunt vol balance adjustment), the ambience was far better and less forced.

So, if anyone is 9ineterested, here is an alternative way to do a final check on the distance between left and right channels, throw a mic at the lp (as long as it is on the centre line and facing forward), do a sweep left and right and compare.

Old hat to a lot I guess, maybe a way to confirm you have it to whatever arbitrary degree of accuracy you think you have.

I'm on a new crusade, 'give me time response over frequency response' haha. Well, I'll take both if I can get it, but time is far more important I think.

 

[RBFC]

This has been an important thread. The minute distances mentioned that can negatively affect imaging are the reason that we should exercise extreme care in our setup protocol.

Lee

 

[vinylphilemag]

This has been an important thread. The minute distances mentioned that can negatively affect imaging are the reason that we should exercise extreme care in our setup protocol.

Lee

Agreed. It has also, for me at least, been a fascinating thread.

 

[DonH50]

Actually, 360 degrees per complete cycle, so about 1.4 deg.

Sorry, everything I did was "per side" so I did the same for that calculation -- 1.4 deg total shift, 0.7 deg/side, so I only used half the circle. I should have stated 1.4 deg total phase shift, sorry.

Both time and frequency response are important, imo. One reason I like RPlusD is that it uses a combination noise/impulse test signal so it's easy to get both from one sweep. It does yield "noisy" impulse output due to the noise signal, but the noise is usually well below the impulse returns we care about.

 

[garylkoh]

Anyway, what prompted me to start the thread (before I found out how wrong I was initially with the physical setup...oh the shame) was how much that timing difference affected things. Sure, it's obvious that it would (and why gary for example stresses in his how to that you need to be down to one mm accuracy) but until you actually experience it you kinda don't know...if you follow.

Terry,
Glad that my paper helped. You do need to get the accuracy down to sub-1mm because a lot of directional information that is processed by the ear/brain interface is in the 1,200Hz to 3,800Hz range. These are the wavelengths that are most distorted by the folds in one's outer ears and gets interpreted as height information.

With most speakers, you will find that the forward-backward tilt (caster angle in car terms) will change the vertical imaging - the vertical height difference between the high-hat and the crash cymbal in a drum set for example. The difference in angle of toe-in/out of the two speakers can change the shape of the soundstage (larger and more diffused on one side, smaller and more focussed on the other). The difference in camber (where the top of the speaker is angled left/right) can twist the stage like a pretzel.

Finally, the distance between the two speakers can change the uniformity of soundstage density across the stage. Too far away and toed-in too much and you have a W soundstage where there is a very dense, focussed center image, a slightly less dense image at the two sides, but the image is diffused between the left speaker and the phantom center, and between the right speaker and the center.

Each and every of the above four can be proven with physics and math plus physiology and psychology of hearing.

I gotta tell you, the difference is quite amazing (hence the thread, I was so 'stunned' by it). As an example, the time difference between the two channels was 200/1 000 000 of a second. Spose I should reduce that, but hey I'm not in school now!! Ok, 2/10 000 of a second difference.

Analog phase shift does not work at all because none of the analog phase shift circuits can shift phase of all frequencies equally. Digital delays are far better, but the processing needed usually means that the delay "jitters" if the clock is not sufficiently accurate. Now we realize why CD/DAC jitter in the range of 10ps to 100ps (picoseconds which is one trillionth of a second) makes such a difference.

As an exercise, when we pass 44100Hz samples to a DSP that has a native clock rate that is a multiple of 48000Hz, what is the error maximum and the minimum error - and what percentage of samples are more than 100ps in error?

And of course the difference did not stop there, the centre image felt far more natural (done time wise-ie properly-than with blunt vol balance adjustment), the ambience was far better and less forced.

We all own systems that cost in the thousands - an investment of less than $200 on a laser rangefinder (do NOT get the cheap laser-pointed sonar ones) - will be the second best ROI. None of us should spend another penny on upgrades or equipment if we don't spend at least a couple of weekends moving our speakers around.

I'm on a new crusade, 'give me time response over frequency response' haha. Well, I'll take both if I can get it, but time is far more important I think.

Which is why we audiophiles hate equalizers because they totally screw with the time information of music. I have a quote that I like to use - the right note at the wrong time is still the wrong note.

This is a FUN thread to be on. I'm glad that so many of the posters also understand physics and math

Cheers!!

 

[terryj]

Terry,
Glad that my paper helped.

It did. I was 'questioning' myself because to seemed to me unbelieveable a) how much the imgae was off centre from 'such a small' difference...except it was not a small difference, oh I'm trying to keep my head up over that one and b) again, I could not believe how much vol difference it required to centre it in the face of that time difference.

The examples you gave in your paper 'proved' to me I was not imagining things.

You do need to get the accuracy down to sub-1mm because a lot of directional information that is processed by the ear/brain interface is in the 1,200Hz to 3,800Hz range. These are the wavelengths that are most distorted by the folds in one's outer ears and gets interpreted as height information.

Maybe you can clarify, but for better or worse I have as the stable datum that 'phase and time' ARE the cues that give us the realistic ambience, the depth and height, the envelopment if you will. And, in genres outside of 'pure minimalist acoustic' music, eg studio based electronic music as an example, then it *must* be the phase cues etc that give the swirling, the sounds that move out and back or revolve around yopur head yada yada?? I mean they all MUST be created in the DAW no? If so, then those effects surely can only come from the manipulation of the phase etc in the signal.

(just on that, I have often wondered how much of that type of stuff is accidental or intentioned. IF they wanted 'that sound' to come out and go around your head, they muast have been able to track it on their monitors etc no?? Which begs the question, how good are the monitors used in comarison to some of the systems here. I think you all can decipher what I mean!! Oh, and another thing, take someone like massive attack. We all bemoan how poorly some recordings are...'the lowest common denominator', stuff like that. OTOH, we have someone like MA and you gotta ask yourself 'what percentage of fans would EVER hear it reproduced as well as it can be reproduced?? It's gotta be a vanishingly small percentage...so you wonder what is in it for the artist to do such a good recording)

With most speakers, you will find that the forward-backward tilt (caster angle in car terms) will change the vertical imaging - the vertical height difference between the high-hat and the crash cymbal in a drum set for example. The difference in angle of toe-in/out of the two speakers can change the shape of the soundstage (larger and more diffused on one side, smaller and more focussed on the other). The difference in camber (where the top of the speaker is angled left/right) can twist the stage like a pretzel.

Yes, I have to get onto that. For now, I just set up a proper centre line this time, made sure the toe in was the same between each, you know, did it the right way this time! THEN I can get on to playing with tilting.

Analog phase shift does not work at all because none of the analog phase shift circuits can shift phase of all frequencies equally. Digital delays are far better, but the processing needed usually means that the delay "jitters" if the clock is not sufficiently accurate. Now we realize why CD/DAC jitter in the range of 10ps to 100ps (picoseconds which is one trillionth of a second) makes such a difference.

The deqx delay is a digital one. Not that it matters. I only really mentioned it because it allowed me very easily to just click on the remote button and compare instantly the effects of havingthe speakers slightly misaligned, to my limit of the deqx (0.02 ms BTW). A lot faster and easier than phycially moving the speaker between auditions. Being able to do that made it so real and clear how important this (FREE) area of tweaking can be.

Others will just have to read your paper and take your word for it.

I do.

This is a FUN thread to be on. I'm glad that so many of the posters also understand physics and math

Cheers!!

I'm glad you were here. It certainly helped me that you were.

 

[DonH50]

I have a quote that I like to use - the right note at the wrong time is still the wrong note.

I use that all the time in my orchestra. Hopefully only during rehearsals! I also tell the conductor all the notes are the same color... It doesn't work.

Reset my speakers today; they were actually as close as I could measure before, but I was moving thing around a bit. Have to do more listening tonight and see what difference I notice. Listening to a couple of CDs today I vacillated between "it made a big difference" and "I don't hear any change". I am not a good test case because it has been too long since I listened critically to a lot of my CDs and I have little time to listen during the week.

There are analog all-pass circuits used to compensate phase shift or add wide band linear time delay, but I have rarely if ever seen them used in an analog EQ. They do add circuitry (and thus noise, power, and cost). It is fairly easy to do with digital filters, however, and linear-phase filters are routinely used in audio (and radar) circuits. I gave up on analog EQ in my home system ages ago because the response aberrations (amplitude and time/phase) across the bands were just too great...

 

[RBFC]

The most useful point made in this thread, IMO, is that both the objective and subjective camps converged on the "truth". Distance and placement measurements, previously thought to be adequate, were mathematically shown to be inadequately precise for producing audibly satisfactory results to critical listeners. Because the science and subsequent listening experience showed image shift with very small placement errors, it seems that all camps (the anal subjectivists and anal objectivists) are IN AGREEMENT! This is hopefully a precedent that we can continue to build upon, by scientifically supporting "subjectivist" beliefs that end up being proven correct by application of solid physics-based calculations. Here, simple trigonometry and acoustic calculations, performed with care and interest, arrived at some very interesting data!

Thank you all, guys. This has been a very gratifying experience and I am learning a great deal.

Lee

 

[cjfrbw]

Well, I am glad the objectivists and subjectivists were able to get their anuses together on something.

 

[RBFC]

Well, I am glad the objectivists and subjectivists were able to get their anuses together on something.

This isn't that type of forum, Carl!!

 

[garylkoh]

Now that we are all in agreement, let's go listen to some music

I tweaked my wife's kitchen system, and I'm now listening to The Very Best of Mantovani on JVC XRCD.

The old Decca recordings has some of the best strings on record.... and this XRCD ain't bad at all.

 

[vinylphilemag]

The most useful point made in this thread, IMO, is that both the objective and subjective camps converged on the "truth". Distance and placement measurements, previously thought to be adequate, were mathematically shown to be inadequately precise for producing audibly satisfactory results to critical listeners. Because the science and subsequent listening experience showed image shift with very small placement errors, it seems that all camps (the anal subjectivists and anal objectivists) are IN AGREEMENT!

We are! Of course, our friends in the objectivist camp will want to tell us on a scale of 1-10 (units to be agreed upon at a later date) how much we agree, while us observationalists will wax lyrical for pages describing how much we agree.

 

[RBFC]

• Post #1. As long as the harp from the pas de deux in Swan Lake blooms out of its corner, the bass drum in Gnomus rattles my innards, the opening string sweeps in Bridge's The Sea float effortlessly into the room, or Muddy Waters appears magically between my speakers, it's all good!
  
  At this point I'm prepared to accept that as long as we all love music, we'll keep approaching those lofty ideals we've talked about for years.

• Post #2. As long as the laser pointer gives me agreement on my speaker distance measurements to an acceptable 5 significant digits, the RT60 of my room is less than my IQ, the highest bass antinode in my room is the bump in the cushion of my chair, and my hearing remains at a level of someone 20 years younger, it's all good.
  
  At this point I'm prepared to accept that as long as we measure carefully, everything will sound its best.

CHOOSE EITHER POST OR BOTH!

Thanks to all.

Lee

 

[terryj]

the RT60 of my room is less than my IQ

dang.

 

[RBFC]

dang.

I still have hope. After all, IQ is just a measurement and doesn't reflect how my brain performs......

Lee

 

[terryj]

hey, thanks! Not only did you cheer me up, I think I just became a subjectivist!!!

wow, the two camps are closer than we thought haha

 

[RBFC]

hey, thanks! Not only did you cheer me up, I think I just became a subjectivist!!!

wow, the two camps are closer than we thought haha

We're all after the same goal.

Lee

 

[flez007]

In my case, I have swapped from obj tu sub just as much as going from tubes to SS again and again - very educational reading by the way... I invested almost 1 hour reading it from head to toe (my wife says I'm ready for Ritalin....)