LQ & OSD: an acoustic alternative

[Elias]

Hello,

If image is locked to the tweeter, possible causes can be:
* too high direct to reflection ratio of the tweeter
* too high level of treble compared to midrange


I understand your center tweeter operates above 7kHz.

Try eliminating (reducing) the direct sound. Why there should be a direct sound above 7kHz at all? Too much of direct sound at this high freqs will not be a benefit in a stereo system because of pinna.

Maybe try ceiling firing tweeter to maintain overall tonal balance and get reduction of direct sound.

Or simply try shelving low pass filter for the tweeter to lower the level, but I think modifying direct to reflection ratio should give better results.

- Elias

Have been listening more and more, I ran into a dilemma

Close positioned or now mono tweeter has a strong point of spaciousness, especially in the depth. Images seem always appear behind the speakers. The presentation is more of 'you are there', instead of 'they are here' -- a 'tradition' of high efficiency horn system (what I had before).

Images in the soundstage are well defined, individual instruments or vocals have their right positions or even shapes and sizes. And I got a feeling that sources seem to have 'haloes' or they're slightly glowing -- there's soft and tender 'atmasphere' around each of them and among them. In addition, such wonder can be maintained in a very large listening area. In extreme off center listening positions -- at the side of the speaker, imaging collapses (of course) while tonal balance stay very stable.

I'm not sure how the images can be projected and focused beyond the tweeters (no special signal process here, only ordinary 1st order filters). Those sounds coming with very high frequecy overtones (attacks of percussions, strings...) are indeed formed far outside of the tweeters when call for, but with a condition...

---- the sound must have its own 'body' in relatively lower frequencies (the fundamentals).

When the sound is purely HF, then comes the problem. Jazz piano trio is one of my favorite forms of music. The drum set is often arranged at one side of the stage (-- almost always on the right in one of my favorit trio, Keith Jarret Trio). I can 'see' the piano, bass, drums at thier correct positions, but when the cymbals are hit, almost always, they appear around the center -- detached from the whole drum set.

Sometimes, a large ride cymbal is softly hit (or hit by a stick with felt ball), there's larger proportion of lower fundamental. This note would come back to its correct position with the whole drum set. Such effect is annoying, be it CD or DVD.

Sigh~ What a pity !

By the way, I made the supertweeter in vertical face to face 'quasi-omni'. And I've tried arranging the midtweeter horns to side or up firing (like the ears of Shrek). Shrek type radiation makes a very spacious sound field, but is not helping the 'positioning' problem of the HF images. And it's too much 'atmosphere' with too less 'body'.

I've also tried "R-L" and "R-L" signal on the midtweeters by series connecting. The sound seems clearer, but could not solve that problem, either.

So, it's not perfect yet. The journey goes on.

 

[CLS]

...

I've also tried "R-L" and "R-L" signal on the midtweeters by series connecting.

....

Oh, I meant 'R-L' and 'L-R'. (Edit time is over... ) You guys must know it.

And thanks a lot for replies.

--------------------

Radugazon,

You've mentioned those test tracks to me earlier. I've downloaded them already, and am now finding a way to put them on to the player... (I'm old fashion my PC is not connected to the audio system .... )

And as mentioned, except for cymbals, (or triangle... ) etc., all other 'normal' instruments are all fine (or excellent) in imaging. Their fundamentals (tend to be ITD) indeed do the job brilliantly.

Those 'pure HF' instruments I brought up have relatively very less proportion of lower frequency fundamental. They give mostly the leading edge of attack. ( No? ) Hmmm.... so that leads to ....


Elias,

Thanks a lot for your suggestions. I've been reading a lot of your posts, very educational and informative. You must know a lot of things I don't know.

In this case, I would tend to agree more on the 1st point (too high direct to reflection ratio of the tweeter), but with uncertainty or confusion.

(Too high level of treble is never an issue on my system, I hate overly bright balance. I just don't let that happen.)

My place is very reflective to start with - concret walls, tile floor, calcium silicate board (and low) ceiling. Absorbing material is almost non-existent.

I've tried 3 types of (super)tweeter setups up to now:

1) Ordinary type, direct pointing at me - the highest D/R ratio, indeed the worst. I took them apart in a short time.

2) 2 tweeters face to face vertically, a fiddled radial type 'quasi-omni'. It gives very uniform HF at the horizontal plane.

3) 45-degree uptilted mono tweeter, Stereolith type. It's more or less off-axis all the time (for almost all possible listening positions), and projecting the most ceiling reflection.


The second type has very little vertical dispersion, its horizontal refections arrive quite long after the direct sound - the difference in path length is > 5m (>14~15ms).

In addition, although the boudaries are all very reflective, the HF sound (especially > 7~8kHz) is intrinsically very 'lossy' along its propogating process. Let alone the multiple reflections. Long delay and high loss, I feel those reflections don't contribute much to the overall 'sound'. It's the 'atmosphere' instead.

Nevertheless, it's the most natural sounding setup. I like it the most, except one drawback - it's oh so quiet, needs too much EQ boost (thus other issues...) Again, I don't like bright sounding. It's the face to face configuration kills the efficiency. I boost it conservatively, and it's measured > 8dB down compared to the mid band. I don't feel comfortable boosting it more....

The 3rd (Stereolith) type gives strong ceiling (early) reflections. By roughly calculation, the path length difference is about 1m or slightly more, which leads to a (critical) 2~3mS of delay. In such HF, TBH, I can't detect any 'smear' or something like that by ears.

The effect of spaciousness is the most obvious in the group. Somehow, sometimes, I feel it's trying too hard. Fascinating, yes, but not always realistic.

-----------

Well, OK, all above are only for supertweeter only (>7~8kHz). I'm more than happy to make this portion mono as I like very much how it sounds to me.

But under that, down to about 1~2kHz, under the principle of OSD, I found the image positioning problem can't be overlooked in my system. Up or side firing setup for this range doesn't solve the problem, even the direct/reflect ratio is lower. I can't get that wonder in supertweeter in this range. Hmmm... isn't this range still in ILD 'territory'? (Maybe this portion of sound is bound up with the 'body' of the instrument much more than the higher range? hehe...)

Or maybe it's because the dispersion patterns are far from ideal (I can't make them omni for the time being.)

I've tried moving the midtweeter (2~8kHz) back to their normal stereo positions. This solved the positioning problem of those HF instruments, but lost a lot of OSD magic. (sure, it's not OSD anymore).

Tradeoffs and dilemmas, as always.

 

[CLS]

A Compromise

I tried relocating the midtweeters to the (inner) sides of big mid horns. See attached picture. (the boxed mini monitors beside TV are not in the main system, they are for TV only)

And the supertweeters are now 2 units (in stereo connection), spaced by 35cm or so, firing almost vertically up with a little angle toward the front. Large proportion of indirect sound in this range is indeed very good. I like it a lot and can't go back. (Thank you, Elias.)

I spent some time trying the toe-in angles and found a compromise among image focucing/positioning, spaciousness, listening area, overall in room response... etc. Now it's kind of an all-rounder, or I should say a compromise.

You may see in the pic that the midtweeters are toed in a lot. I suppose the objects nearby might probably cause some problems. I'll try some absorbing materials later. Or bringing the midtweeter higher. Or both...

It's alright for now. To be continued for sure....

 

[ScottG]

I tried relocating the midtweeters to the (inner) sides of big mid horns. See attached picture. (the boxed mini monitors beside TV are not in the main system, they are for TV only)

And the supertweeters are now 2 units (in stereo connection), spaced by 35cm or so, firing almost vertically up with a little angle toward the front. Large proportion of indirect sound in this range is indeed very good. I like it a lot and can't go back. (Thank you, Elias.)

I spent some time trying the toe-in angles and found a compromise among image focucing/positioning, spaciousness, listening area, overall in room response... etc. Now it's kind of an all-rounder, or I should say a compromise.

You may see in the pic that the midtweeters are toed in a lot. I suppose the objects nearby might probably cause some problems. I'll try some absorbing materials later. Or bringing the midtweeter higher. Or both...

It's alright for now. To be continued for sure....

Have you tried it with the mid-tweeters very close together but "toed-out"?

 

[Radugazon]

Toe out : do you think of something like this ?

 

[ScottG]

Toe out : do you think of something like this ?

YES.. though for his waveguides that is too much toe-out. (..and his won't have directionality due to dipole cancellation.)

BTW, very nice work Radugazon! It's far closer to a perfect radiation loudspeaker system then just about any I've seen (..and I've never said that before).

Have you been convolving your media to binaural?

 

[graaf]

Have you been convolving your media to binaural?

uhm uhm... tricky question

 

[ScottG]

uhm uhm... tricky question

..it is after all the final ingredient to unlocking a correct stereo perspective with this setup.

 

[Radugazon]

Good questions are always tricky...

Before answering, I have to say that here the system is set up in a side firing configuration, but of course, except this, it's not a stereolith. It's composed of linear quadrupoles with their usual positive and negative properties. The subs remain dipoles and are very excentered.

more details if needed :

This has the potentiality of creating a huge spaciousness, but at the price of a serious imaging. As I am a pragmatic guy, I use some diffusers to correct this (see the picture...).
OMG, they are not quadratic...It's randomly sized bamboos. Following the diameter, I suppose they must work over 6000 Hz and reflect a part of the spectrum on 180°.
They finally narrow the image too a realistic size on a little formation, but keep it wide on a big forte. In any case it preserves the repartition of the sources. But you know, all this is subjective stuff.

Then, their orientation on the floor horizontal plane has little effect, but the vertical orientation is the key:

oblique inside, it widens, oblique outside it narrows...

An other way of controlling the "center channel" is precisely to toe out the HF group....

I have diffusers like this close to the ceiling too, all weird, but it works.

 

[Radugazon]

Now, pre-convolving all my records for satisfying an always temporary setup is not a good plan. I have some samples I did months ago, they are now unlistenable, all like with added reverb and echoes.

I would like better a generalist system that "transforms" (bye bye the concept that you know ) an ordinary multi tracks record in something convolved-like, without inducing trouble in the naturally or artificially binauralized ones.

It's not a joke but don't take my rhetoric at 100%, the subject can't be contained in two lines.

Here the subjective approach has a good helper : the phase analysis, I mean the stereo phase between each channel, not the phase between the drivers of an isolated speaker.

On the joined pic, a binaural record of a 1000 Hz signal with 180° rotation of the field taken from the listening spot (5 meters).

Normally, when it's like this, the system images rather more accurately than ultra wide. Work are still in progress, I try to conciliate both aesthetics. Many factors have some importance, as the foam phase plugs that change a lot of things (thx Mr Geddes).

Last word: on a rotating phase test, a classical basic system produces a Right/Left alternance, but in a system like this one, or like the stereolith, it's easy to perceive (or imagine...image/imagination) a rotating source around the system.

3D ? Definitively yes.

 

[CLS]

Hi Scott,

I've breifly tried toe-out, or even side firing. Unfortunately that didn't work properly in my place. Ah~ in fact I don't know what 'proper' is. Anyway toe-out or side-firing was not good to my ears (and mind)...

In the pic of post #43 you may see a drape on the left corner. There's a big pillar behind that. While the space on the right is open. So if the refections form both sides are stronger than a point (in proportion), the images would be pulled to the left too much. OTOH, large toe-in setup sidesteps that problem ( -- side reflections are largely delayed, attenuated, and appearing on the opposite sides... )

I thought of diffusers, too. But it's not easy to get a good balance among all dilemmas in my case. There's always practical reason, sigh~

In the experiment of up-firing tweeters, I got a feeling that omni might be the way to go (-- or very close to omni, like Radugazon's LQ...). Now I'm working on it.

 

[LineArray]

...

Recently, I’ve put the system in a classic configuration (takes 2 hours only) that produces almost square waves and that’s average in off axis dispersion. This has some qualities, but it’s not playing music, just HiFi. Before I was biased on it, but now, it’s really kicked out.

...

Hello Radugazon

i like that bit from you, that i have set in boldface ...
yes, obvoiusly there is a difference to some of us.

How did your "classic configuration" look like ?

Did it also have the midrange LQ configuration ?

 

[Radugazon]

This is an old picture, still with temporary boards on the sides, but basically it's the same architecture. This one has the domes facing the listener. No LQ here, it's all dipoles (but you know that dipole in the highs is an other story...)

for the almost square waves, it's really "almost", but it's the best I had. Just an other transcription of the step.

 

[LineArray]

Ah, i see ...


OK, i noticed you and some others being involved in OSD and
also analysing the "stereolith".

My comment is not on that, but on your LQ midrange
configuration posted in "Sound of the Jungle" ...

By using that midrange LQ you managed to get a sound source
which radiates a virtually spherical pattern.

But there is only one frequency - given perfect drivers - where
the spherical pattern is perfect and the wave front emanating is
coherent, right ?

That 'adjustment frequency' or 'mid frequency' is the one, where
the distance is right. For wideband opereration, the distance has
to be tuned to a 'mid frequency' as a compromise, as i understood
your description.

For differing frequencies, there will be an interference pattern, which
may maintain the gross spherical dispersion to some extent, but the
coherence is lost and the phase you get at a single point in space is
somewhat 'random'.

That behaviour from the midrange LQ in itself
- diffusivity and wideband spherical radiation pattern - may contribute
to mitigate early (room) reflections correlated to the direct sound.
The system sounds more 'airy', there is an increase in 'depth',
the speakers are less localizeable as sound sources themselves.

You feel that tonality is not changing much when moving around
in the room, even when outside the stereo listening zone ?

Spaceousness remains, even when localization of phantom sources
may get fuzzy for listenening places in the periphery ?

The recording venue - as a tendency - seems to dominate the
listening room, which (better than with classical conf.) may disappear ?

---
I guess your midrange LQ as a "diffuse like omni" makes a large contribution
in making that kind of 'music'.

It is not only from the overall OSD configuration, i am convinced.

But please, go on ... you're doing admirable things ...


btw.

http://www.stereotimes.com/images/mbl101d_black_180x495.jpg

I always doubted, that especially the MBL 101 'big midrange' unit radiates
a coherent spherical pattern. But i might never have the chance to analyse
the vibrational behaviour of that unit, so just guessing.

What i want to say is, that you may have 'modeled' more and even important
aspects of that trancducer, which seems to have inspired you, than you might
have noticed. And even if my assumptions on MBL 101 should be incorrect,
you have introduced an interesting property, which has value in itself regarding
speaker/room interaction.

 

[Radugazon]

Explaining things to somebody that understand better than me is challenging . .Consider then it's some real life opinions.


I've tried a lot of LQ variations, bare drivers, bigger baffles, great an small
inter-driver space, cones inside, cones outside, radial, horizontal (actual
config) and even oblique. I put them in concurrence too with the dipole wiring.

the LQ has at least two big negative points :

• absolutely distorted impulse that has a lot of preringing

• high passed system, one octave higher than the corresponding dipole

Then, when comparing it to the theoric model, the famous recessive inverted lobe only exists in the inter driver space with good impulse and no high pass.

The outside lobes dominate.

other unknown points :

• as you say, there's a pivot frequency that is optimal for a given spacement.

• how wide is the optimal band ?

• what relation with the spacement?

• what about the resulting interferences? Combing ?

the only sure fact is that the sound is pretty dead when the drivers are very close.

now the positive:

• If we forget that the drivers are not perfect and have some one axis beaming, the LQ is an amazing beast to measure: always the same curve shape, anywhere could be the mike (except very close to the walls). Still more evident for the room response measurement.

• "seen" at 90°, they are difficult to localize, but the SPL is here. This kind of system listened on one channel only already fills the room on one side.

• the reflection pattern is for sure a diffuse one, chaotic.

as you see it's almost a complete confirmation of your deductions, except for the first paragraph.


-------------------------------------------------------------

Comparing radial LQ vs Horizontal allows a better understanding of each formula

• the horizontal excites more the walls, it over widens and distorts the image if no low pass applied (the center is then very weakened)

• the radial must excite more ceiling ahd floor, but this remains unoticeable.

• the tonality remains the same

this proves that finally this setup is not perfectly omni if it's bandwidth is too wide.

BTW, with the diffusers, the image gets more narrowed but can extend much more with the horizontal LQ.
Depth is the same,looks like the imaging is accurate for me (difficult subject).

But the horizontal LQ renders better a 360° rotating signal.

both of them can be listened laterally,some one can turn the head or even plug one hear, the general localization remains and it's hard to localize a speaker.

but the priveligiated seat is really on axis and at 5 meters.

Is this system faithful with the recording venues? I have some very contrasted records, so it makes the difference, but the room has it's personality superposed. Difficult to investigate a bias.
This makes everything bigger only IF there is enough SPL. I listen mostly at a level that would be forbidden if any neighbors. Listening at low levels makes it cheap and slim. Bias again.

---------------------------------------------------------------

This is for the mids, but they are not alone. I think that an innocent listener would notice more the bass and the sparkle of the HF, as always...the mids are the forgotten relative.

----------------------------------------------------------------

long question, long answer...