CD² (Constant Directivity and Critical Distance)

[Y~BNA]

Frequency response just up to 45deg horizontally does not tell much about overall directivity, particularly with a diaphragm of that size. You can expect it to narrow down above 5K horizontally due to the diaphragm width, which is a bit too early to use is fullrange. If you want to keep lobing and overly increasing directivity index under control, I would assume that a single unit might go up to 7K, but that already requires a pretty clever strategy to implement a slimmer (super-)tweeter without vertical lobing.

Combining it with a curved or truncated line array equipped with identical units below 2K should be doable, so you can keep the directivity constant between 500 and 5,000Hz. Given how tall the diaphragm area is, I would say 3 or 5 units are ideal, with the central one being active to approx 5 to 7K, and the outer ones having some low-pass filtering applied around 2K or a bit lower.

There are examples of this planar used in arrays, for example on diyaudio
https://www.diyaudio.com/community/...ght-line-array-floor-to-ceiling.347227/page-3

Would say it is a very challenging project but also a promising one. Have heard a ton of speakers equipped with such linesource-shaped planar tweeters, and noticed a tendency that these combined with usual midrange cones kind of appeared to be disappointing, while those combining such planars with midrange line arrays or dipoles, were always surprising in a positive way.

Hey Arindal, not sure i understand you correctly. By 'combining' you mean 'implementing it as', correct?

I am indeed planning that, but not shading from the center to the outer units, but instead shading from the center of the speaker as a whole, that is, where lowest unit meets the midwoofer it is crossed over to. That one will be at 0dB and each unit above tapered down an incremental step 1~1,5dB upward to 5~7,5dB attenuation on the topmost unit of the six (need close to 1200mm of line source to meet Crtical Distance aims). As shown below.
But, to make this conversation somewhat easier, i finally have some preliminary sketches to share, so folks can have an iimage that speakes louder than a thsand words, as the saying goes...

 

[Arindal]

By 'combining' you mean 'implementing it as', correct?

I referred to the term ´combining´, if we are talking about a concept of pairing a midrange line array of any type plus a single line driver for the treble (which can physically be the same driver, even an 1.5-way array with only the central driver playing fullrange, is imagineable).

In case you opt for a full-range (above 500Hz) array with all drivers getting the same range, of course you might want to describe it as ´implementation´. Given the specific planar driver you have mentioned, or the geometry of an array in which it is actually is fitting (particularly the spacing), I see a lot of reasons why it should not be used full-range, not even as a conventional shaded array.

That one will be at 0dB and each unit above tapered down an incremental step 1~1,5dB upward to 5~7,5dB attenuation on the topmost unit of the six (need close to 1200mm of line source to meet Crtical Distance aims)

A sketch would really help here, and without a simulation, it is not really clear how this array will work with smaller wavelengths. An array, shaded solely step by step in amplitude per driver, particularly a non-symmetrical design as you have proposed, bears all sorts of risks the higher you go in frequency. Vertical directivity narrowing down quickly, lobing, phasiness, combfiltering effects, you name it. Operating a single planar driver above a certain frequency, solves the majority of these issues while bringing the concept closer to constant directivity at the same time.

Speaking of CD: You might want to achieve line source properties at lower frequencies with a 120cm array, but that leaves the question open what happens below crossover point and which frequency that will actually be - a sudden step to omnidirectional pattern would not be helpful for achieving constant directivity, particularly not for crossover points above 400Hz.

All the best for your project!

 

[Y~BNA]

Hey y'all!

Some preliminary sketches of the basic setup. Dimensions yet to be adapted for optimum acoustical trade-offs.
(Such as, in this depiction, Sub baffles will interfere with Woof radiation, Sub baffle/position must shrink/move, Woof raised by a bit, or both, etc. still solving the dimensional puzzle).

The speaker will generally be for larger rooms, but can be used as a single unit when left and right section are attached and toed-out in medium rooms, with a physical channel divider (as per D.B. Keele) then applied. The section from 650~670 and up ('upper body') can be swivelled from -23 to +23 degrees for toe-in or -out, so user can experiment a bit and find best config for their room. Both standing and sitting listeners should get a full frequency spectrum and good intelligibility between about 1,5m and 8m in any not overly reverberant room.

It resembles a four-way (3+sub) in which MID and HI are joined in a dipole line-source array of planar Mid-Tweeters (~670Hz to ~18kHz), with a dipole Woofer from 80~90 to 650~670Hz, and a pair of back-to-back summed dipole Subs below that. So a four-way disguised as a three-way, tx to GRS PT6816. An admittedly low spectrum brand, but these planars have been tested and found exceptionally well behaved!
I do not have great concerns, as mentioned by a few, of narrow radiaton pattern on these planars. The radiating area of the tweeters is not much wider than 22mm, which would be enough to keep them omni up to some 7kHz without issue, but then they have a grid, pressing the waves through vertically stacked arrays of 6mm wide openings. The grids were not invented by GRS but copied from several reputable forebears of this charmingly priced new product.
If Huygens-Fresnel principle is worth the paper it's written on, this would signify new, independent sources of propagation that, at less than 1/3 of 18kHz's 19mm wavelength, will introduce a wide spreading diffraction even when not in a line array.

But they are, on top of that, in a 1170mm (~4ft.) line array, which according to Griffin's meta-research paper 'Design Guidelines for Practical Near Field Line Arrays' (2003), keeps the near field quite large (~4m@2kHz) and expanding steeply above that. Between 650Hz and 1.5~2kHz the array may lack a bit, with CD being around 2~3m, which i will try to address by a not-to-steep XO from the Woofer.

The array behavior is further gonna be tuned slightly, tilting it back a small angle (6~10 degrees) and shading the amplitude, from 0dB at the unit next to the Woofer in 1~1.5dB steps to -5~-7.5dB at the uppermost unit. Precedence effect is expected to help smearing, especially in the XO, and improve transients, and relieve slight lobing above 8kHz a bit. Spacing of the units @37mm between effective radiation areas will keep them well behaved up to almost 10kHz, and also this is somewhat optimized by the physical and amplitudinal shading.
This could also improve above-mentioned near field imperfection by directing the main lobe at the sitting and standing listener when closer to the speaker.

All sub/woofers X up more than an octave below calculated Dipole Peak and several octaves below rated highest operating frequency. Woofer also X down an octave above rated lowest operating frequency.
Tweeter XO at 670Hz is about half an octave above it's first measured anomaly on the low end, where LR4 will render acoustic energy of said anomaly down 18dB (-6dB@XO+12dB for half octave = ~1.6% remaining energy) if i'm calculating rightly. Will be tuned flat before XO implementation if measured to be necessary.
Their rated response starts at 400Hz. Combined with the fact six units share a 4.5 octaves where only a small fraction of the voltage lands, i feel confident a (possibly asymmetrical) 650~670Hz XO will not present any issues for the planars.

The 670Hz incidentally happens to be the frequency at which woofer arrays' critical distance falls off to way within four meters (Griffin, 2003), which is however, according to same meta-research paper, mitigated by in-room responses. Hence, a good transition frequency away from line the array? In any case, the sub/woofer collection in this setup would seem equally capable of using said in-room responses to compensate LF behavior, possibly even without the drawbacks of the boxed woofer arrays all research was based on, up to 2003.
Research included far field and near field methods to calculate and test array responses, by many reputable authorities, including Urban, Heil, Ureda, Lipschitz&Vanderkooy and some ten more, mostly around '00 to '02 but starting as early as '64, '86, '90, and '97, extracted and combined by the late James R. Griffin Ph.D. (rip 2024). Recommended resource for living room line arrays!

Anyway. Shoot me now! Bury me later... and if directed to literature supporting your shots, that will of course be doubly appreciated!

 

[Arindal]

It basically is a four-way in which MID and HI are joined in a dipole line source driver array of planar mid-tweeters (670Hz to 20kHz), with a dipole woofer from approx. 80 to 670, and a pair of dipole SUBs below that. So a four-way disguised as a three-way tx to GRS PT6816.

Okay that answers at least some of the questions. I would call it a 3-way, crossover points at 80 and 670Hz, open-baffle dipole concepts for sub and midbass, dipole-line-array-hybrid for mid-tweeters.

these planars have been tested and found exceptionally well behaved!

Have never measured one myself, but I have seen some THD measurements. While they do behave as intended and go much lower than other planars, I would first take a unit (or two or three stacked to a short line array), apply necessary DSP for linear anechoic response, and do some THD measurements under true baffle-less conditions. The do show some increase in THD below 2K, even more under 700Hz, and in full dipole mode you loose a significant amount of SPL to dipole cancelation at larger wavelengths. Your concept might push them beyond the red line, or not, only measurements at elevated SPL can show.

Same testing method might be required for dipole woofers. You did not mention the intended drivers, but full baffle-less operation at lower frequencies might demand an unexpected amount of power and excursion capability. I have occasioned dipole concepts which did not quite behave as their inventors had been hoping for, so strong recommendation for linearization and then THD measurement.

I do not have great concerns, as mentioned by a few, of narrow radiaton pattern on the planars. The radiating area of the tweeters is not much wider than 22mm, which would be enough to keep them omni up to some 7kHz without issue, but then they have a grid, pressing the waves through vertically stacked arrays of 6mm wide openings.

That does not help much to prevent them from narrowing horizontal radiation pattern. The datasheet suggests an effectively active diaphragm area of 38mm in width, horizontal listening window will narrow down earlier than expected. Not critically alone, but maybe too steep to achieve constant directivity, as the vertical directivity of any tall planar driver, particularly with several units stacked at some distance between each other, will either this or that way increase steeply, starting at pretty low frequencies.

If you manage to get around the difficulties I was mentioning (which might require a totally different solution for the treble, like an additional smaller tweeter), this can be a truly great concept. Keep on rocking!

 

[Y~BNA]

Thanks for your systematic approach (and your time of course).

Okay that answers at least some of the questions. I would call it a 3-way, crossover points at 80 and 670Hz, open-baffle dipole concepts for sub and midbass, dipole-line-array-hybrid for mid-tweeters.

Still sleeping on what o call 'em, but definitely the least of my worries!

Have never measured one myself, but I have seen some THD measurements. While they do behave as intended and go much lower than other planars, I would first take a unit (or two or three stacked to a short line array), apply necessary DSP for linear anechoic response, and do some THD measurements under true baffle-less conditions. The do show some increase in THD below 2K, even more under 700Hz, and in full dipole mode you loose a significant amount of SPL to dipole cancelation at larger wavelengths. Your concept might push them beyond the red line, or not, only measurements at elevated SPL can show.

Am gonna measure the planar lines separately of course, before deciding on XO points and slopes. But they ARE intended as dipoles, two of 'em are supposed to illuminate a room, so i'm guessing twelve of 'em could illmnate a room with less stress and good line array characteristics. The radiation patern of which, theoretically, is not that far from the fgre of eight in LF point source dipoles. I AM gonna delve into your observations though. Griffin (informedly) comments on some of 'em, but it's getting close to crash time for me, so i'll bust your balls later.

Same testing method might be required for dipole woofers. You did not mention the intended drivers, but full baffle-less operation at lower frequencies might demand an unexpected amount of power and excursion capability. I have occasioned dipole concepts which did not quite behave as their inventors had been hoping for, so strong recommendation for linearization and then THD measurement.

There's no full baffle-less implemenation. There are small baffles, with a distance D (as per Linkwitz) of close to 170 for the Midwoofers and a D around 270 for the Subs, where the latter also have the floor and one sidepanel as "infinite" resp. "extended" framelength D, around almost half of their circumference. That would do something for their efficiency i expect. The sizes, Xmax, Vd, sensitivity, etc. of all woofers are indicated in the sketches. X-linear of the 18" moves about 4 liters per stereo channel. The midwoofer about 0.36 liter. You think that will not fill a room? They don't have to be deafening, they have to sound beautiful at realistic levels.

That does not help much to prevent them from narrowing horizontal radiation pattern. The datasheet suggests an effectively active diaphragm area of 38mm in width,

which is partially damped by a felt inlay. The actively vibrating width is really some 22mm.

horizontal listening window will narrow down earlier than expected. Not critically alone, but maybe too steep to achieve constant directivity, as the vertical directivity of any tall planar driver, particularly with several units stacked at some distance between each other, will either this or that way increase steeply, starting at pretty low frequencies.

The vertical directivity should be limited, floor and ceiling highs are like pride. It only hurts, never helps (Pulp Fiction). No? I depend on the line arrays to have narrow vertical dispersion, as predicted, and wider than a single unit horizontal, as also predicted. The driver to driver gaps do not cause significant lobing if kept down below 25% of total in height (check), and shaded (check), according to the literature. But will still test...

If you manage to get around the difficulties I was mentioning (which might require a totally different solution for the treble, like an additional smaller tweeter), this can be a truly great concept. Keep on rocking!

Tx so much, more later, cheers!

 

[Arindal]

two of 'em are supposed to illuminate a room, so i'm guessing twelve of 'em could illmnate a room with less stress and good line array characteristics.

As mentioned, if you drive them into increasing acoustic short circuit by true dipole arrangement and low crossover freq and correcting for that one, reserves might get eaten up quickly. Just mentioning it in order to take precautions.

Having heard quite some hybrid dipoles which were surprisingly good, I noticed that they all had pretty chunky dynamic midrange cones deployed with lots of excursion reserves (Linkwitzlabs, Ecouton, Spatial).

X-linear of the 18" moves about 4 liters per stereo channel. The midwoofer about 0.36 liter. You think that will not fill a room?

Do not mean to say it will not fill the room, and I am more concerned about the smaller midwoofer. But it is worth an experiment including THD measurements and listening tests, how it will behave in its lowest octave. With sufficient power, it might work, but there is a certain risk it leads to increased THD while sounding subjectively thin.

which is partially damped by a felt inlay. The actively vibrating width is really some 22mm.

Is there felt on both sides, or just the rear? I would not be sure this has been implement for a full damping hence limiting the active are for broader radiation, which is unlikely to happen at such wavelengths (5-8K). My guess would be the felt rather serves a purpose of reducing resonance issues.

The vertical directivity should be limited, floor and ceiling highs are like pride.

That is essentially the idea of a line source, which should work perfectly well in your case for the midrange. But if you take closer look at the angles which are causing ceiling and floor reflections, they should not be too far off from tonal balance and constant vertical directivity as well. A line source of 120cm height might be causing overly narrow listening window and significantly reduced energy in the treble region, while a dipole midwoofer does produce a lot of midrange energy below 670Hz under angles which cause these reflections. Doing something to get this reflection window closer to constant directivity, might be a good idea.

The driver to driver gaps do not cause significant lobing if kept down below 25% of total in height (check), and shaded (check), according to the literature.

As mentioned, it might work with these tall planars, but there are also chances it might cause overdamping of treble in the room as well as the effect of lobing in terms of cancellation, (matt treble, ´lack of air´, as audiophiles would put it), or even some phaseyness.

 

[Y~BNA]

As mentioned, if you drive them into increasing acoustic short circuit by true dipole arrangement and low crossover freq and correcting for that one, reserves might get eaten up quickly. Just mentioning it in order to take precautions.

I will defnitely check if their XO in the 650-670 region will put too much stress on them, especially if SPL compensation for DP fall-off is at issue. Check.

Having heard quite some hybrid dipoles which were surprisingly good, I noticed that they all had pretty chunky dynamic midrange cones deployed with lots of excursion reserves (Linkwitzlabs, Ecouton, Spatial).

The mid is quite chunky at 12", not? And it does not have to worry about too much highs, crossed from (perhaps instead of 80~90, better) 100Hz to 670Hz. But that too will only be seen at testing. It IS a dedicated OB design, with 7mm of linear movement.

Do not mean to say it will not fill the room, and I am more concerned about the smaller midwoofer. But it is worth an experiment including THD measurements and listening tests, how it will behave in its lowest octave. With sufficient power, it might work, but there is a certain risk it leads to increased THD while sounding subjectively thin.

Is there felt on both sides, or just the rear? I would not be sure this has been implement for a full damping hence limiting the active are for broader radiation, which is unlikely to happen at such wavelengths (5-8K). My guess would be the felt rather serves a purpose of reducing resonance issues.

You're right, just the rear, so for resonance isses indeed. Yet, it does limit the free movement of the membrane. But they come so well recommended, i can't wait to connect them and see what characteristics they truely have. Hopefully in a week or two i'm ready to find out. Will share.

That is essentially the idea of a line source, which should work perfectly well in your case for the midrange. But if you take closer look at the angles which are causing ceiling and floor reflections, they should not be too far off from tonal balance and constant vertical directivity as well. A line source of 120cm height might be causing overly narrow listening window and significantly reduced energy in the treble region, while a dipole midwoofer does produce a lot of midrange energy below 670Hz under angles which cause these reflections. Doing something to get this reflection window closer to constant directivity, might be a good idea.

What would you suggest to do? Of course i first have to test how the 12" and the line source will cooperate in real life, but what could be done?

As mentioned, it might work with these tall planars, but there are also chances it might cause overdamping of treble in the room as well as the effect of lobing in terms of cancellation, (matt treble, ´lack of air´, as audiophiles would put it), or even some phaseyness.

All the intel i can find on the planned setup predicts that the shading and the tilting and the truncated height of not even 1200mm have no real need to misbehave, but i'll admit i'm still nervos about it. Anyway, nough said, i'll supply some hard data ASAP.

 

[Y~BNA]

Btw, the woofers are planned to be mounted as shown in below schematic image, which is not immediately apparent from the front and side views above.
Also here the dimensions will not be what you see, especially the distance between the back to back will be more, to make the cavity less tricky, and better air flow.
But basically very shallow H-frames is the principle, for the midwoofer adapted as a wave guide front, and bevelled back for air flow improvement.
The subs are not compound/isobaric, it is two woofers back to back, summing with a fase shift of some 10% of wavelength at highest F (~100Hz, -30dB 200Hz) decreasing to a theoretic 2% at lowest 20~25Hz, 'thickening' the radiation pattern slightly, but giving some more headroom to the lowest 2 octaves.

 

[Y~BNA]

Have you looked at the factory data sheet? It shows the off-axis response being in the same ballpark as a 1" dome tweeter, albeit closer to constant-directivity in the top octave than would be typical for a 1" dome. Anyway, the pattern narrows north of 4 kHz or so by much more than one might expect from a single 6mm-wide aperture.

I will try to find a way to have them radiate wider, either in the way they are mounted, possibly combined with a waveguide-resembling baffle/mounting rod, or by the way the loudspeakers are directed in the room. I am specifically prepeparing to make them suitable for central/toe-out placement, because i built that once and got an unexpected seriously good result in articulation and soundstage imaging.

On the other hand imo the horizontal radiation pattern looks a bit wider and more uniform than one might expect from the planar diaphragm without the apertures, so I think they are making a worthwhile improvement in the radiation pattern.

Theory would suggest that. Huygens-Fresnel principle

I think your goals are good and your overall game plan is good, and even if you don't achieve all of your goals perfectly I think you'll still end up with a really sweet system.

Tx, that s good advice also.

Don't get overly caught up in what the Directivity Index number is; imo what you want is a long enough time delay between the arrival of the direct sound and the strong onset of reflections so that they don't perceptually "fuse". And imo 10 milliseconds is long enough (though more would be better).

I think (depending on F) 10ms is close arousuperimpose another reverb over thatnd the bare minimum for psycho-acoustically (subconsciously) distinguishing the direct and reverb sound as separate sonic events. But indeed from about that delay and up, reverb is not detrimental by definition, and sometimes even a good thing.
Some folks uphold that anything but direct sound is detrimental, as in anechoic is the best room. I would know how to defend that, for instance, the reverb of the venue (be it a hall, studio, whatever) is already in the recording, so superimposing another layer of different reverb over that could only be detrimental.
Yet i think our psycho-acoustic processor would notice something is off when listening in an anechoic room, getting ONLY the direct souund form loudspeakers.
This is a very interesting topic to me, and of great imiportance to speaker design IMO.

I do have a couple of thoughts, and if they're not useful just ignore them:

First, I think you will definitely need waveguides on your stack of planar drivers to avoid having far too much early reflection energy off the same-side walls. I suggest making your best estimate of how much toe-in you can get away with, and then choosing your waveguided radiation pattern width such that you are not strongly "illuminating" that same-side-wall. Imo it would be desirable for your first strong horizontal-plane reflections to be the long-path, across-the-room reflections off the OPPOSITE side wall. So I'm thinking your target radiation pattern width might end up being in the range of 60 to 90 degrees wide, depending on the specifics of your room and speaker placement options..

Interesting, since this is precisely what my central toe-out setup also addresses. And i will tell you this: the opposite wall reflects left reverb to the right ear and vce versa. The centre positioned near field monitor i mentoned above, was feeding right direct sound as well as right reverb as first arriving to right ear and left to left. Which, without hard evidence other then listening, i suspect of being responsible for the unexpected great sound that little thing gave.
Anyway, the section from 670Hz to 18k in this project will be flexible: you can swivel it toe-in to toe-out from approx. +23 degrees to -23 degrees.

Second, what are your plans for the backwave?

They have just two rows of radiating grid-openings instead of four, on the back. First i'm gonna mount them as a stack. Then gonna see what the frot and rear patterns are, and choose which is preferred to serve as front in my setup. The difference in pattern is much, i'll try to adapt one or both.

If you are able to position the speakers at least 5 feet out from the front wall, you can probably let the planar array's backside disperse as wide as it pleases. That being said, imo there is an argument for waveguiding the backwave as well, such that when you "aim" the front wave you are simultaneously "aiming" the backwave.

In my Spaish place i'll have all the room in the world to place them 3/6/9/12 ft. from the front wall, the side walls, etc. In Amsterdam i'll probably set 'em up in somebody else's home!

I wish I had a solution for the vertical spacing issue.

I have come across several reputable sources stating the vertical gaps between line source drivers in a truncated array are not a serious issue from about 2m (6~7ft.) and up, if you meet a few parameters: the gaps don't occupy more than 25% of the vertical size, the array is ampltude shaded, and possibly bit physically shaded (tilt rearward). I check all three.

You might make some triangular cardboard cut-outs representing different radiation pattern widths and set them down on the floor where the speakers will go and play around with different "toe-in" angles, to help you visualize what it will take to avoid strong early same-side-wall reflections. Imo THOSE are the enemy that would thwart your plans.

Yes. Which is one of the reasons Siggy was so happy with the figure of 8. So i'm gonna attempt as much figre of 8 as i can. Almost ready to start dong shit. In some two weeks hope to share first preliminary impressions. Tx for your input, very appreciated.

 

[Y~BNA]

You can only make a speaker “room-independent” to any meaningful degree by making its directivity extremely narrow from roughly 80 hz and up. But that’s not some universally 'ideal' speaker, it’s a speaker built around one very specific goal.

And that goal comes with its own compromises. The real engineering move isn't trying to delete the room or chase an abstract ideal, it's figuring out what the design actually needs to do, then choosing a concept that hits those requirements while keeping the trade-offs under control.

One of the first requirements you have to nail down is directivity: how wide do you want the beamwidth to be? Then comes the second most important question: what max SPL are you aiming for?

Well, what my speaker actually needs to do is make the room less important, which could be as valid an engineering goal as any, no?
I'm not sure i agree with your statement, although you definitely have valid points.

But, innocent like the day i was born, i was quoting Siggy, who also expained how his dipoles were a considerable step to the solution of that challenge.
Over the whole range where the dipole fig-8 pattern is upheld, early side wall reflections are diminished (good), reflections are dominated by the late ones (good), and in bass range reverb is 4.8dB down compared to a monopole with the same on-axis SPL. (good).

This does not make THE ideal speaker, nothing does as you rightly state, but it is a signficant step. Before he passed away, he mentioned living room line arrays tickled his curiosity. They do mine too.

I can however, for myself, list a number of things that would make a speaker usable for many people of different creeds, in many different homes with different acoustic properties.

-Large near field, small far field.
-Much direct sound, little reverb (same thing to an extent).
-As little early reflections as possible.
-Close to even spectral content over distance and direction.
-Line array -3dB per doubling D for low frequencies (can be addressed differently).
-Dipole fig-8 for high frequencies (can be addressed by wide spreading tweeters in array).
-Great clarity at SPL bearable for mortals just chilling.

Some of those are difficult or impossible, some can be approached, and several can be combned in a single design. That would be a very fuctional design for people who love music but do not spend their lives sitting in chairs. I'm one o those, so my design choices are for my fellow all-over-the-place dwellers.

Can you help me phrase that goal? Or have you advice that could us me closer? You would be one of my hero's!

 

[Blockader]

Well, what my speaker actually needs to do is make the room less important, which could be as valid an engineering goal as any, no?
I'm not sure i agree with your statement, although you definitely have valid points.

But, innocent like the day i was born, i was quoting Siggy, who also expained how his dipoles were a considerable step to the solution of that challenge.
Over the whole range where the dipole fig-8 pattern is upheld, early side wall reflections are diminished (good), reflections are dominated by the late ones (good), and in bass range reverb is 4.8dB down compared to a monopole with the same on-axis SPL. (good).

This does not make THE ideal speaker, nothing does as you rightly state, but it is a signficant step. Before he passed away, he mentioned living room line arrays tickled his curiosity. They do mine too.

I can however, for myself, list a number of things that would make a speaker usable for many people of different creeds, in many different homes with different acoustic properties.

-Large near field, small far field.
-Much direct sound, little reverb (same thing to an extent).
-As little early reflections as possible.
-Close to even spectral content over distance and direction.
-Line array -3dB per doubling D for low frequencies (can be addressed differently).
-Dipole fig-8 for high frequencies (can be addressed by wide spreading tweeters in array).
-Great clarity at SPL bearable for mortals just chilling.

Some of those are difficult or impossible, some can be approached, and several can be combned in a single design. That would be a very fuctional design for people who love music but do not spend their lives sitting in chairs. I'm one o those, so my design choices are for my fellow all-over-the-place dwellers.

Can you help me phrase that goal? Or have you advice that could us me closer? You would be one of my hero's!

One speaker that actually achieves incredibly narrow beamwidth, satisfying that "much direct sound, little reverb" and ''As little early reflections as possible'' requirement, is the Reflector Audio Square Two. It is a horn design where the woofers surrounding the horn mouth act as a beamforming array, shaving off the excess dispersion. It has its own cardioid bass extension(that extension box is only available in Russia). It can easily reach 110db continuous SPL thanks to having a HF1440 compression driver for mids and HF.

To achieve incredibly narrow dispersion, there are essentially two approaches:
Horn + beamforming: This is what Reflector Audio's Square Two does.

Parabolic speakers: Meyer Sound designed one capable of 160 db SPL. To my knowledge, this is the only example of a parabolic speaker in commercial use.

looks scary. If you want to minimize reflections as much as possible, you do not have any other choice than these. Some of your requirements eliminate 99.9% of the available speakers out there.

-Dipole fig-8 for high frequencies (can be addressed by wide spreading tweeters in array).
-Line array -3dB per doubling D for low frequencies (can be addressed differently).

I have no idea, what you are trying to solve with these requirements.

Large near field, small far field.

This doesn't make sense.

Now, if you want extremely narrow bass like ±15 (30 degrees total dispersion), the way to do that is with second order gradient designs. The catch? These only exist practically in microphones. To pull it off with speakers, you'd need something like four extremely capable 21" drivers just to hit 90 dB at 20 Hz. Good luck with that.

 

[Duke]

I think (depending on F) 10ms is close...

My understanding is that a 10 milliseconds interval is enough time to give us that differentiation down to about 700 Hz, which is "low enough" to give good results.

Yet i think our psycho-acoustic processor would notice something is off when listening in an anechoic room, getting ONLY the direct souund form loudspeakers.
This is a very interesting topic to me, and of great imiportance to speaker design IMO.

Agreed.

... this is precisely what my central toe-out setup also addresses. And i will tell you this: the opposite wall reflects left reverb to the right ear and vce versa.

It's not obvious to me that toe-out does this. I would expect toe-out to more strongly illuminate the same-side wall while more weakly illuminating the opposite-side wall. What am I missing?

 

[Arindal]

The mid is quite chunky at 12", not? And it does not have to worry about too much highs, crossed from (perhaps instead of 80~90, better) 100Hz to 670Hz.

A single 12" in de facto baffleless design is not much for 100Hz.

But I was more concerned about how the planners behave between 500 and 1000Hz. It might work, but I would really doublecheck if THD is not overly increasing due to the planars working in acoustic short circuit mode.

What would you suggest to do? Of course i first have to test how the 12" and the line source will cooperate in real life, but what could be done?

To ensure a relatively even vertical behavior of the line (leading to a more constant directivity) and a smoother transition towards the dipole midwoofer, you could for example use a 3-way segmented/shaded line array consisting of 5 drivers with different low-pass filters.

670-1,200Hz - all 5 planars
670-2,500Hz - the inner 3 planars
670-20,000Hz - just the central tweeter active

An elegant solution which does not change the general design, just needs alterations in terms of crossover design.

Another variant would be to implement a midrange line array instead of a single 12", but that would basically mean a complete redesign.

i was quoting Siggy, who also expained how his dipoles were a considerable step to the solution of that challenge.

Having heard his LS521 in its latest iteration on several occasions, I would assume that these are indeed an exemplary solution when it comes to tonality and room-induced bass and lower midrange problems. In terms of localization at greater listening distances, and making them completely independent from placement and room reverb, rather not.

Coming back to your original post, my hypothesis would be, he had prioritized constant directivity over sufficiently high directivity index which is crucial for practical listening distance. Which might absolutely work in certain rooms with some restraints concerning placement.

Before he passed away, he mentioned living room line arrays tickled his curiosity. They do mine too.

I wholeheartedly agree, but can say from own experience that this is a very challenging field of loudspeaker design with very little reliable research and publications from the home audio perspective. Particularly when combining it with dipoles, which is just the second principle that IMHO defies predictions how it will sound. Encourage you to do some field testing of existing products, as companies like Lyngdorf, Perlisten, KSD, IO Designs and MEG have offered pretty promising solutions in recent years.

It is a horn design where the woofers surrounding the horn mouth act as a beamforming array, shaving off the excess dispersion.

Combining different strategies of controlling directivity, one in front of the other, ist from theoretical point a very promising design idea, and might turn out being the holy grail of loudspeaker design in terms of directivity and critical listening distance. I would personally neither use a two-dimensional cone array nor a horn, but, hey, solutions might just differ.

 

[Y~BNA]

My understanding is that a 10 milliseconds interval is enough time to give us that differentiation down to about 700 Hz, which is "low enough" to give good results.



Agreed.



It's not obvious to me that toe-out does this. I would expect toe-out to more strongly illuminate the same-side wall while more weakly illuminating the opposite-side wall. What am I missing?

One. The speaker stands central, is further from the walls, which is lacking in most peoples setups. That makes large portion of early reflections landing late, passing you by, mostly reaching you via the long path (rear wall). Reason has it (angle of incidence is angle of reflection) this adds to a preferable late to early reflecton ratio.

Two. Contrary to what i was always told, this positioning does not (or did not in my experiment) harm the sound stage. To be honest, subjectively listening so far, i had the impression it improved sound stage imaging, if anything. I suspect because the right and left horizotal side reflections come from the same side as the left and right direct sound, increasing channel separation/decreasng cross talk, whatever the term is.

Anyway, my design up to now is modular: the two speakers can be one unit placed centrally (with physical separation panel mounted if desired), or be two separate speakers placed traditionally. In both applications the 700-20k section can be swivelled IN as well as OUT at will (+23 to -23 degrees or so), enabling experiment with best angle adjustment on the fly.

 

[Y~BNA]

One speaker that actually achieves incredibly narrow beamwidth, satisfying that "much direct sound, little reverb" and ''As little early reflections as possible'' requirement, is the Reflector Audio Square Two. It is a horn design where the woofers surrounding the horn mouth act as a beamforming array, shaving off the excess dispersion. It has its own cardioid bass extension(that extension box is only available in Russia). It can easily reach 110db continuous SPL thanks to having a HF1440 compression driver for mids and HF.

That sounds like an experiment i would also wanna do. But the narrow beam limits the direct sound to the narrow space within it. The late to early reflection ratio would seem to improve in this approach, but the nearfield behavior would become too focused for my goals.

To achieve incredibly narrow dispersion, there are essentially two approaches:
Horn + beamforming: This is what Reflector Audio's Square Two does.

Parabolic speakers: Meyer Sound designed one capable of 160 db SPL. To my knowledge, this is the only example of a parabolic speaker in commercial use.
View attachment 507525

looks scary. If you want to minimize reflections as much as possible, you do not have any other choice than these. Some of your requirements eliminate 99.9% of the available speakers out there.

Wow, does that work in a home setting? How big is it? Is it fit for acoustically small spaces?
Btw, 160dB close up can kill a person! At bass frequency your gut and organs are ripped apart.

I have no idea, what you are trying to solve with these requirements.

Like i said, i don't spend my life in a chair, i'm all over he place doing whatever. So i want near field behavour spread about the room, even if it is a large room.

This doesn't make sense.

Large near field small far field doesn't make sense to you? You want critical distance small, great sound at one sweet spot in a chair nailed to the ground?

Some speakers have less room interaction than others. There's various design principles that apply features improving near field behavior over wider areas. Both OB and LA are potentially useful in that sense, though one lacks in higher voice and above range, while the other lacks in lower voice and below range. I'm reading everything i can find (including Merlijn van Veen at Meyer Sound) to see if we can tame LA for the living room.

Now, if you want extremely narrow bass like ±15 (30 degrees total dispersion), the way to do that is with second order gradient designs. The catch? These only exist practically in microphones. To pull it off with speakers, you'd need something like four extremely capable 21" drivers just to hit 90 dB at 20 Hz. Good luck with that.

Neh, narrow is not where i wanna go, though your horn suggestion did get me thinking.

But OB has 4.8dB less reflected low bass energy, compared to a monopole at same on-axis SPL. That's reflected bass 75% reduced. That is a great step toward direct sound dominating the in room field.

Tx very much for your extensive input and ideas. Even though we're on different wavelengths in some topics, it's still very thoughtful input. Good research is not finding evidence your idea's are correct, but just the opposite: if you search for evidence they are incorrect, and after much effort you can't find any, THAN you're on the right path. So i'm not there yet and you help me stay cautious.

 

[Y~BNA]

A single 12" in de facto baffleless design is not much for 100Hz.

Maybe not, i'll be very cautious with it and possibly exchange it for another long throw model, or put a second one back to back like i'm doing below 100Hz.

But I was more concerned about how the planners behave between 500 and 1000Hz. It might work, but I would really doublecheck if THD is not overly increasing due to the planars working in acoustic short circuit mode.

I'm a bit worried about that too (which is healthy), but at 6dB more sensitivity than mid i'm hoping to just drive 'em at very low V per unit for now. If not, i'll need to make this a four way and have not one but two XO in the holy 300-3000 area.

To ensure a relatively even vertical behavior of the line (leading to a more constant directivity) and a smoother transition towards the dipole midwoofer, you could for example use a 3-way segmented/shaded line array consisting of 5 drivers with different low-pass filters.

670-1,200Hz - all 5 planars
670-2,500Hz - the inner 3 planars
670-20,000Hz - just the central tweeter active

An elegant solution which does not change the general design, just needs alterations in terms of crossover design.

Which brings me to this indeed elegant idea, that has to caveats:
1. I have a three way amplifier/DSP unit per side available. For this i would need 5-way, meaning getting a 2x6 separate DSP unit, five amps and two SMPS's per stereo side. That is for another day.
2. The line array is physically adjacent to it's aforementioned woofer brother, also for XO purposes, which n your proposal would put the central two (there's six, not five) planars at about 80cm centre to centre from said brother.
I am therefore planning to keep the first planar (approx. 29cm c-c at present) at 0dB, and each next one 1 of 1,5 dB down. If 1dB, the top planar will then be shaded 5dB, and the point between the woofer and the lowest planar will be at ear height when sitting close to the speaker, while the physical shading (backward tilt) will be so slight that you can still sit further away. And standing can also be close and further away. Bass and low voice will be the limiting factor for near field behavior further away.

Another variant would be to implement a midrange line array instead of a single 12", but that would basically mean a complete redesign.

Yessss i'm already dreaming of my CBT dipole or stuff like that, but this is gonna research OBLA hybrid concepts. So i'm really needing to use both to do what they (can) do best: DP to manage the bass modes and deliver CD into the lower voice, and LA to deliver CD in the higher voice and keep higher voice and treble at equal levels while maintaining optimum direct/reverb ratios.

Having heard his LS521 in its latest iteration on several occasions, I would assume that these are indeed an exemplary solution when it comes to tonality and room-induced bass and lower midrange problems. In terms of localization at greater listening distances, and making them completely independent from placement and room reverb, rather not.

Because the dipole behavior is no longer doing it's job from higher voice and up. That's where my hybrid idea is coming from. And LA's are no longer LA's when reaching just about my proposed XO. That's why i'm so happy i came across planar tweeters that can be crossed as low as 600Hz.

Coming back to your original post, my hypothesis would be, he had prioritized constant directivity over sufficiently high directivity index which is crucial for practical listening distance. Which might absolutely work in certain rooms with some restraints concerning placement.

Yeah, he was a tinkerer. When done tinkering, he would sit. In one spot. Between his speakers. In a rather absorptive room.

I wholeheartedly agree, but can say from own experience that this is a very challenging field of loudspeaker design with very little reliable research and publications from the home audio perspective.

Griffin collected some LA research approached from the living room perspectve. Since most has been directed at PA applications, based on far field calculating, and is not relevant to acoustically small rooms and near field. Maybe you should check him out.
But Keele (whom i think Griffin did not include, i forget) is also a briliant audio engneer. He turned 87 the month before last. He also used both physical and gain shading going up (not from the middle, you know what his CBTs look like). The sound stage is said to be BEHIND his CBTs. Never heard 'em, alas.
And there's some interesting posts and tests and images (with measurements) of OB LA builds by Perry Marshall on DIYaudio.com. Not hybrid like mine, full OB-LA.

Particularly when combining it with dipoles, which is just the second principle that IMHO defies predictions how it will sound. Encourage you to do some field testing of existing products, as companies like Lyngdorf, Perlisten, KSD, IO Designs and MEG have offered pretty promising solutions in recent years.

And buying all kinds o speaker cabinets? Right, want my account nr. so you can donate? I'll mention ur name in the credits. Share 10% of profits. But i have sniffed around a few of those, without ever hearing them though.

Combining different strategies of controlling directivity, one in front of the other, ist from theoretical point a very promising design idea, and might turn out being the holy grail of loudspeaker design in terms of directivity and critical listening distance. I would personally neither use a two-dimensional cone array nor a horn, but, hey, solutions might just differ.

Stop putting even more deas in my head!!!
Neh, seriously, your ideas and suggestions and serious considering has helped a lot, make me stop and think, etc. Appreciated. In about a week or two i'll have some news. If Sonarworks still has the calibration file of my 10yo. XRef20 (never been used), the news might even be supported by evidence.

 

[Arindal]

If not, i'll need to make this a four way and have not one but two XO in the holy 300-3000 area.

Would not worry about x-over points if the drivers are close to each other. And why 4 way? A midwoofer line array would do the job of filling the gap between the dipole sub and the planar array.

I have a three way amplifier/DSP unit per side available. For this i would need 5-way, meaning getting a 2x6 separate DSP unit, five amps and two SMPS's per stereo side. That is for another day.

A passive first oder low-pass filter and everything driven by one amp channel, would do. It is basically two inductors (like in the region of .8 and 1.5mH), and maybe an additional resistor (6.2Ohms) for the central fullrange planar to compensate for a second planar in serial operation. Not urging you to do it like this, but maybe you can keep the possibility open and invest 30 bucks into inductors later.

And buying all kinds o speaker cabinets?

Was more referring to listening to demos at dealerships or visiting some trade fair.

All the best for your project!

 

[Y~BNA]

Would not worry about x-over points if the drivers are close to each other. And why 4 way? A midwoofer line array would do the job of filling the gap between the dipole sub and the planar array.



A passive first oder low-pass filter and everything driven by one amp channel, would do. It is basically two inductors (like in the region of .8 and 1.5mH), and maybe an additional resistor (6.2Ohms) for the central fullrange planar to compensate for a second planar in serial operation. Not urging you to do it like this, but maybe you can keep the possibility open and invest 30 bucks into inductors later.



Was more referring to listening to demos at dealerships or visiting some trade fair.

All the best for your project!

You've been a great support with ideas, considerations, advice, etc. And very thoughtful. I'm keeping your mid line array proposal open, since redesigning the construction is my specialty. Designing the physics, the acoustics, is my learning curve in this project. So if measurements confirm your concerns to be real issues, i may stll do that. Tx for everything! Will post again after measuring.