An Improved Array

I was thinking of the steering effect in the straight array and not using an out of phase driver on the front end, just wondering out loud if that's the beam steering they used as they have about 3 big modes, selfish (where they optimise for one listener), normal, where the beam width would be used to supply a couch and party where it would behave more like an Omni.

Now these 3 modes aren't their real names, but you get the drift.

B&O would be able to change the phase and level of each driver. That's a lot of freedom right there.
However it may just be level and delay. Though I expect they do use some extra tricks in the selfish mode with mid/side EQ.

I'm still trying to process how this thing works, but here's some thoughts:

1) We can't get too aggressive with delay or EQ, because both will affect the directivity. If you look at the hornresp sims, you'll notice that the wavefront shape is almost perfectly cardioid. So if you were to EQ one of the drivers differently than the other, that would change the directivity pattern.

This is something that I learned with cardioid midbass arrays in my car. When they work, they're unbelievable. But the 'trick' is that each driver in the array has to be equalized to have the same response shape. IE, if you had four drivers in an array, and one driver was 6dB louder at 100Hz than another driver in the array, then the directivity would change at that frequency. The trick is to keep the response shape of each driver identical across their bandwidth. Note that this DOESN'T mean that they have to play at the same SPL level; the curve itself just needs to have the same shape. Varying the SPL level of the elements in the array can be used to steer the wavefront.

Naturally, this can be used to our advantage. For instance, if the measured pattern wasn't perfect, EQ could be used to 'smooth out' the polar response.
narrow1.png

That may explain how they're getting these epically smooth polar curves. Admittedly, waveguides can get close to this. But as noted a few posts back, waveguides don't sound the same. (I think this speaker sounds better.)

2) I would speculate that the most powerful tool at our disposal is the shape of the enclosure. If a perfectly hexagonal enclosure yields a cardioid, then what does a pentagon yield? What happens if you elongate the enclosure? (Note that the patent shows a hexagon, but the real speaker appears to be 'squished' in the X axis.)

3) I would also speculate that the second most powerful tool would be varying the intensity level. I *really* wish Hornresp had the ability to vary the output levels in the wavefront simulator.
 
Last edited:
The mid-side EQ I was getting at is for the stereo pair, not on one loudspeaker.
mid/side as in phantom centre and (Left and right).

They use more than one trick to optimise the listening spot.
First of all, as KR described in his review, the BeoLab 90 system uses digital signal processing (DSP) to correct the room's acoustic. However, it differs from conventional correction solutions in applying not just individual filters for the Left and Right speakers, but also filters to correct the speakers' summed (Mono) output and the difference between their outputs (Side). The responses of these filters, as set up by B&O's Geoff Martin, are shown in fig.1 (Mono and Side) and fig.2 (Left and Right). Both sets of filters are applied simultaneously; you can see that the amount of correction is relatively small, covering a range of +5 to –6dB.
Read more at Bang & Olufsen BeoLab 90 loudspeaker Measurements | Stereophile.com
 
I believe I have the Beolab 90 sorted out. I know this is a long and complex thread; if you're going to read one post in this thread, this is it:

First off, why do we want directivity control? I believe that we want directivity control because it reduces the room's influence on the sound of the recording. For instance, with a Danley SH-50, what you hear is what's on the recording. If there's a big soundstage on the recording, you'll hear it. It's true that many speakers can produce a BIG stage, but many that do make *everything* sound big. The SH-50 isn't like that; it's basically an x-ray on a recording.

Second, what's so great about the Beolab 90? It's great because it offers the virtues of something like an SH-50. But it's more attractive, it doesn't use compression drivers, and I find that it combines the rock solid imaging of an SH-50 with the pleasant ambience of a good dipole. The reason is because it offers the directivity control of a waveguide while using conventional drivers.

It *is* possible to get good results with a simple Cardioid like Linkwitz's LX Mini. For instance, in an online review, the LX Mini was described like this:

"Overall, the resulting soundstage from the LXmini presents as a soundstage plane, with exciting width and depth and depth acuity that go far beyond what most speakers can accomplish.

The description above is consistent with what I hear, and the reason that I've enjoyed cardioids like the LX Mini and the Gradient Helsinki. I'd definitely love to hear Juhazi's AinoGradient.

aDYLoaO.jpg

Although the LX Mini sounds very good, it's response curves are not exceptional. This isn't a knock against Linkwitz; there's only so much you can accomplish with a resistive cardioid.

Dipole_speaker.gif

So let's take a minute and think about why this is. In a dipole speaker, we have positive radiation in the forward lobe, out-of-phase radiation to the back, and nulls to the left and the right where the two wavefronts interfere.

lxmini-pair-2.jpg

In the LX Mini, acoustic damping is used to absorb a fraction of that back wave. So we end up with more energy going FORWARD than going backward.

But why does the response of the LX Mini vary so much? The reason is because acoustic damping varies with frequency. IE, six inches of damping will have a larger effect on a wavelength that's three inches long than a wavelength that's 30 inches long. Due to that, a resistive cardioid will never have the same directivity across it's bandwidth.

5d4eAYE.jpg

This is an 'ideal' cardioid response pattern.

6iNgtyr.png

This is the predicted response pattern of a hexagonal loudspeaker enclosure, similar to the array in the Beolab 90. In the predictions, you can see that it's possible to get very close to cardioid response, particularly at high frequency.

The reason why it's easier to get cardioid response at high frequency than low is because the enclosure needs to get bigger and bigger as the wavelengths get longer. For instance, if you want directivity control down to 1000Hz, you need an enclosure that's approximately 34cm in width. (13.5")

r520zLX.jpg


Although the midrange array in the Beolab is under a foot wide, the measurements indicate that directivity control goes all the way down to 200Hz. So what gives?

BTjujxV.jpg

qNtSDsl.jpg

I believe the 'trick' is that it is also an end-fire array, but a very interesting one. I believe it is a end-fire array but only at certain frequencies.

This next part makes my head hurt, so I hope that I'm explaining this correctly:

If you look at the sims of the hexagonal array, you'll notice that the wavefront shape is almost perfectly cardiod at high frequency. As the wavelengths get longer, the directivity control begins to collapse.

The concept behind a end-fire array is very simple. In an endfire array, we put our speakers in a line. Then we delay the speaker at the front so that the array becomes directional.

b3CjZio.jpg

axDyyMv.jpg

Here's a comparison between one speaker's radiation (spherical) and an end-fire array of twelve. Obviously, the end fire array is far more directional, at the expense of efficiency. (Thanks to Peter Morris for the sims.)




Okay, if you've made it this far, congratulations. In summary we have a hexagonal loudspeaker which produced a cardioid wavefront, particularly at high frequency. To make the speaker more directional than it's small size would indicate, an end-fire array is also used.

But we're not done yet.

I'd hoped that the end-fire array would work across the board. But it doesn't. The reason that the end fire array doesn't work across the board is because it creates severe lobes, particularly at high frequency.

So we have a conundrum here. We have one thing that works well at high frequency (the hexagonal enclosure) and another thing that works well at low frequency (the end-fire array.)

qNtSDsl.jpg

If we look at the Beolab 90 from overhead, we can see why the end-fire isn't a cure all. For instance, 1000Hz is 34cm long. If we delay the midranges at the FRONT of the enclosure so that the wavefronts from the BACK of the enclosure line up in FRONT of the enclosure, we'll get a strong lobe facing forward of the speaker. That's exactly what we want, that's one of the reasons we're getting directivity control without the use of a waveguide. But at 500Hz, that same delay will generate a null on-axis. I hope that makes sense; the delay that improves our directivity at 1000hz will screw up our frequency response at 500Hz.

The solution is simple: we need a variable delay. For instance, if we have seven midranges cover the frequency band of 350Hz - 2500Hz, we want virtually no delay at 2500hz, but at 350Hz we want as much as 1-2ms of delay. The net result is to reinforce the forward lobe at low frequency (where the enclosure is too small to do it) but not to screw up the polar response at higher frequencies, where the enclosure is physically large enough to control directivity on it's own.
 
The reason I linked the article in post 20 is that is describes how BeoLab 90 achieves its beamwidth control:

...

"The total result is a careful balancing act where some things are known or given:
  • the frequency range of the music being played through a given loudspeaker driver (this is limited by the crossover)
  • the natural directivity patterns of the loudspeaker drivers on the loudspeaker enclosure within that frequency range
  • the locations of the loudspeaker drivers in three-dimensional space
  • the orientations of the loudspeaker drivers in three-dimensional space (in other words, which way they’re pointing)
Note that these last two have been calculated and optimised based on a combination of the natural directivity patterns of the loudspeaker drivers and the desired beam widths we want to make available.
As a result each individual loudspeaker driver gets its own customised filter that controls
  • the level of the signal at any given frequency
  • the phase (which includes delay and polarity – sort of…) of the signal at any given frequency
By controlling the individual output levels and phases of each loudspeaker driver at each frequency it produces, we can change the overall level of the combined signals from all the loudspeaker drivers in a given direction. If we want to be loud at 0º (on-axis) and 20 dB quieter at 90º (to the side), we just apply the filters to all of the drivers to make that happen. If we want the loudspeaker to be only 10 dB down at 90º, then this just means changing to a different set of filters. This can be done independently at different frequencies – with the end goal to make all frequencies behave more similarly, as I talked about in this posting and this posting."

Also, since the filters are merely settings of the DSP (the digital signal processing “brain” of the loudspeaker), we can change the beam width of the loudspeaker merely by loading different filters – one for each loudspeaker driver in the system.

The end result is a loudspeaker that can play many different roles, as was shown in the different plots in this posting. In one mode, the beam width can be set to “narrow” for situations where you have one chair, and no friends and you want the ultimate “sweet spot” experience for an hour or two. In another mode, the beam width can be set to “wide”, resulting in a directivity that is very much like an improved version of the wide dispersion of the BeoLab 5. In yet another mode, the beam width can be set to “omni”, sending sound in all directions, making a kind of “party mode”.
 
Oh I've read that already :)

I'm going to CES next week, had some time to kill due to the holidays, thought I'd reverse engineer it.

IMHO, B&O can't possibly be making much money on the Beolab 90. The cost of the drivers is something like $6000, the cost of the amplifiers is in that ball park, the cost of the DSP also. The enclosure is a solid chunk of aluminum.

So either this is a showboat designed to show off what they can do, or they're going to take technology from the B90 and trickle it down to something more affordable.

If it's the latter maybe I can come up with something that doesn't require 38 channels of DSP for a stereo pair.
 
Thank you Patrick for the brainstorm! It would be nice to know excactly what 90 does when settings are changed, I believe that both delay and spl are changed.
Judging from Stereophile's measurements, it is a 4-way system
117Beo90fig5.jpg


I believe that 4-way is needed for wide-range directivity control, and dsp/multidriver tricks help a lot to achieve that specially in low and low-mid registers. Variable directivity patterns set another extra challenge for dsp, but that is just a matter of parameters when one has multiple drivers and enough channels available.
 
I'm pretty sure the late John Dunlavy would have considered the STEP to be a 'train wreck'.
I have to add that Stereophile isn't always measuring in the right spot to really 'catch' the impulse as designed.

117Beo90fig5.jpg


It seems they try to make up for it by exaggerating the bottom end:
117Beo90fig3.jpg


I can understand not wanting any latency as this set is supposed to be a play all sources kind of speaker being able to directly hook up to a TV set.

The low end delay might explain this:
Geoff said:
Interestingly, I have certainly heard cases where a good loudspeaker can make a recording sound worse than it is. Specifically, it is not difficult to buy home loudspeakers (or headphones) that have more bass at lower frequency ranges than some studio monitors. As a result, you might hear content or artefacts in the low end that they did not hear in the recording or mastering studios. Had they heard it there, they might have decided to remove the low end problems. A good example of this is Eric Clapton’s “Unplugged” album. This has a very low frequency ringing/rumble in it that sounds very much like the sound of a tapping foot causing a microphone stand to vibrate and shake a microphone. On “normal” loudspeakers, this is inaudible (as I suspect it was in the studios where it was recorded and mastered) but on large loudspeakers with unusually low low-frequency ranges, this is very audible, and very annoying.
Test Tracks: What not to play! – earfluff and eyecandy

I have a pretty strong low end (not as exaggerated), and have it arrive pretty much in time. For me the low rumble sounds add something positive to enhance the "live feel" of that track.
 
Last edited:
I'm pretty sure the late John Dunlavy would have considered the STEP to be a 'train wreck'.
I have to add that Stereophile isn't always measuring in the right spot to really 'catch' the impulse as designed.

117Beo90fig5.jpg


...

Hi Ronald, I agree and kind of surprised to see this as it is no secret what an accurate step response is. Consider an ideal speaker with a frequency response from 20 Hz to 20 kHz with some low frequency roll off as most speakers would not want to pass DC. Here is such an example using 15 Hz as the corner frequency, amplitude top chart, corresponding step bottom chart:

ideal%20XO%20ampl%20and%20time_zpsehwfdfor.jpg


In the step response, one can see that all frequency arrive at the same time (i.e. the vertical "step") and the decay is based on the type of cabinet alignment. Here is an example of 4 types of alignments, still with a 15 Hz corner frequency:

step%20response%20different%20rolloffs_zpsz4m0elpx.jpg


I have time aligned my 2-ways to match the ideal step for my cabinets alignment and very happy with the results: Just purchased a pair of JBL 4722n speakers. - Page 198 - AVS Forum | Home Theater Discussions And Reviews

I am surprised at the BeoLab step response as the speakers are not time coherent. I wonder if that is the sacrifice/tradeoff with respect to beam steering...

Happy holidays, Mitch
 
I think waveguides are the 'gold standard' for directivity control.

The thing that's so enchanting about the Beolab 90 is that it has that 'waveguide speaker' sound, but with more spaciousness.

I'm not sure if I can describe the difference, but I'll try:

Big waveguides eliminate early reflections, and I believe this is a large part of the reason that their imaging is so pin-point. For instance, a Danley SH-50 can create a center image that's so solid, you would swear there's a speaker there. The images are solid as a rock.

Dipoles create a big spacious soundstage, largely due to reflections off the back wall I think. Cardioids are similar in nature. For instance, the Gradient Helsinki creates a very pleasant soundstage, it's big and it's spacious.

Beolab 90 seems to do both; it has that 'giant headphone' effect like the SH-50, but there's also "space" like the Gradient Helsinki.

One theory I have is that we can perceive the early reflections in a waveguide, and it reduces the soundstage size. Basically the bigger the waveguide, and the "less perfect", the more the soundstage suffers.

This is pure speculation of course. About the only evidence that it may be true is that I've noticed that waveguides that feature really clean impulse response seem to "disappear" into the soundstage better. An example of this is the 18Sound XT1086.

I wish I had some big LeCleach horns to test this hypothesis. If you look at their simulated impulse response in Hornresp, it's about as clean as it gets.

But back to the subject at hand!

f3f14edfb3d37dcac4c6ea708c20f070.jpg

The Beolab 90 patent is very difficult to follow, but it's referenced Mr Stiles' Bessel patent. The B90 uses a circular array.

So...

It's possible that the circular array in the B90 is some type of improved Bessel array, possibly inspired by the Stiles array.

prod-anthony-gallo-acoustics.jpg

Here's one last thing to ponder. I don't know how many of you have heard the Gallo Acoustics speakers. They come *really* close to that ideal that I mentioned, the ability to generate a rock-solid center image while also creating a big stage. Basically the best attributes of a waveguide speaker like the SH-50, along with the best attributes of a good cardioid or dipole.

Obviously, the enclosure plays a big part in this.

td-m1_wh_06-100531272-gallery.png

Fujitsu's speaker also has jaw-dropping imaging. At CES last year, it was shocking that this unassuming speaker imaged better than many $100K setups. And look at it! It's basically a $15 Tymphany TG9 in an egg. It sells for $1000.

Just to be sure that I wasn't taking Crazy Pills, I 3D printed a similar enclosure, and loaded it with a Tymphany TC9. Sure enough, it sounds incredible. (Measures really great too!)

Of course there's a "but" here...

The power handling on these speakers is dreadful. Imaging in spades, no dynamics.

But a Bessel array of five drivers raises the output level by 6dB over a single driver, and the improved Bessel array described in Stiles' patent gets that even higher.

When considering directivity, I think it's critical to look at it in 'constant' instead of 'controlled' terms because at the end of the day, power response is what we're after, isn't it?

We've been using waveguides to control directivity lower in frequency to meet up with large woofers who narrow early......while smaller mids and midwoofers don't begin to beam until much higher relative to their diameter.......and the newer BMRs don't seem to beam much at all!

We know that controlling directivity below 800hz is a very difficult task and pretty much out of the realm of typical DIY circles sans the Dipole systems which introduce an entire set of new problems.

What you've observed with the small single element speakers.....I have observed the same in that they have what i 'perceive' to be fantastic imaging!......IF you're in the near/on axis sweet spot. I have a very simple MMTMM build at home using 3" midwoofers crossed to a dome at 4khz on a 4" baffle that images better than any speaker I've ever owned or listened to..........as long as I'm seated on the tweeter axis between them. There's also the very limited dynamic range of course and poor LF dynamics to consider........but I think there's much merit in looking more closely at what these small point sources do so well which begs the question about diffraction. Linkwitz himself can find no subjective reason to consider it in his designs......should we?
 
......and while we're looking at this subjectively, let's consider another speaker that doesn't play the diffraction game yet is heralded for its superior imaging, the Continuum which is modeled after the LS3/5. No rounded edges, a recessed woofer and edges for direct radiators to bounce off all over the place.......yet set them 5ft apart on stands and they literally disappear into a room. There's something to be said for that. Is diffraction a good thing subjectively?.....is confusing the location service of the ear/brain function a neccessary design goal? If so, is there a frequency band where this might be more effective?.......or am I back at the psychoacoustics place?.......

Maybe, but at the end of the day I really don't care about the measures and graphs when I'm sitting down for a good listen!........and if I'm meandering about the house while listening I care even less!

I'll share my greatest listening experience which came in what many consider to be the absolute worst type of listening space and that's the old Hayden Planetarium in NYC.......a massive concrete overhead dome where the system fired up into the dome and yet the entire dome was simply illuminated with sound across the entire spectrum. No placement of instruments or voices in a particular point in space.......they were simply everywhere!.....and yet the effect was the most 3D experience I can remember. I repeated it several times with different laser light shows over the years and the effect was always the same.

And I'm fortunate in my very large shop which has an old beam timber barrel roof to have been able to recreate some of that magic by having our multipoint music service speakers fire up into the barrel. I can walk through the shop or showroom all day and it just sounds wonderful to me........the sense of space......the music all around me........

Maybe that upfiring woofer in the LXMini is doing something in the Fairfield we or Linkwitz hasn't considered?.........hmmmmmmm.
 
We must remember, that useally we are discussing about loudspeakers in small rooms. Radiation pattern/diffractions in free anechoic space is one thing and interaction with boundaries of a room is another.

An interesting but very difficult topic here! Bang&Olufsen RD folks gave us a tour de force example of what can be done. Another much more simple case is Kii Audio Three acoustics
 
If you don't need the variable directivity of the Beolab 90, the Kii solution is probably more "sane." With the Kii solution, you basically have a conventional two way loudspeaker for the midrange and treble, and then a four driver woofer array is wired to produce a cardioid. (It's likely an end-fire array.)

As noted in my post from two days ago, I think the 'secret sauce' is to vary the delay. At the upper limits of the woofer array, the delay will create lobes. So at the upper limits, we use zero delay. At the lower limits, the wavelengths are long enough so that lobing isn't a problem, and then we can use quite a bit of delay.

BTW, this is why the Kii and the Beolab are very deep cabinets. The Beolab is nearly a meter deep. Even with DSP delay, physics still matters, and we can only get pattern control with a big device. If it was a waveguide it would be wide, like the Danley SH-50. Since it's an end-fire array, it has to be deep.
 
......and while we're looking at this subjectively, let's consider another speaker that doesn't play the diffraction game yet is heralded for its superior imaging, the Continuum which is modeled after the LS3/5. No rounded edges, a recessed woofer and edges for direct radiators to bounce off all over the place.......yet set them 5ft apart on stands and they literally disappear into a room. There's something to be said for that. Is diffraction a good thing subjectively?.....is confusing the location service of the ear/brain function a neccessary design goal? If so, is there a frequency band where this might be more effective?.......or am I back at the psychoacoustics place?.......

Maybe, but at the end of the day I really don't care about the measures and graphs when I'm sitting down for a good listen!........and if I'm meandering about the house while listening I care even less!

I'll share my greatest listening experience which came in what many consider to be the absolute worst type of listening space and that's the old Hayden Planetarium in NYC.......a massive concrete overhead dome where the system fired up into the dome and yet the entire dome was simply illuminated with sound across the entire spectrum. No placement of instruments or voices in a particular point in space.......they were simply everywhere!.....and yet the effect was the most 3D experience I can remember. I repeated it several times with different laser light shows over the years and the effect was always the same.

And I'm fortunate in my very large shop which has an old beam timber barrel roof to have been able to recreate some of that magic by having our multipoint music service speakers fire up into the barrel. I can walk through the shop or showroom all day and it just sounds wonderful to me........the sense of space......the music all around me........

Maybe that upfiring woofer in the LXMini is doing something in the Fairfield we or Linkwitz hasn't considered?.........hmmmmmmm.

I'd say there's quite a difference between wanting accurate imaging or the "music everywhere" type of sound. Not saying that last type can't be satisfying or even worthy of a goal all on it's own. I've heard that type of sound in a store using PA equipment cleverly adjusted. It was like walking trough a cloud of music. Much better than any music club, Disco and what not I've ever had the pleasure of visiting.

On the topic of diffraction, sure there are area's in the FR where diffraction might not be very offensive. In stereo there are a few holes in a perfectly setup listening triangle. But when aiming for accurate imaging, I'd keep an eye on diffraction and generally don't care if someone like Linkwitz does not. He also doesn't care about time coherency and I do.

I count myself lucky that I've achieved some of the greatest listening experiences of my life right here in my own home. To achieve it I did (have to) listen to Linkwitz, and Dunlavy, Geddes, Toole and all the rest. I wouldn't even want to exclude Choueiri as he rightfully pointed out some 'real' problems as well. Many more not listed here... My personal goal: A "you are there" sound experience. Quite different from "they are here" or "music everywhere" though it has definite similarities to that last type in the listening area. If all works right you get that same engulfment of music around you, but still can place or point out each source of the sound.

This achievement from B&O is nothing but spectacular. But (if he were still around, RIP) Dunlavy would (or at least could) claim it isn't accurate and he would be right. This does not mean it wouldn't or couldn't sound spectacular. Just not accurate in a strict technical sense.

If Kii kept an eye on timing like Bruno's other project with Grimm Audio it could even be more accurate. As long as it has the balls like these B&O monsters because you would really need that bottom end too!

In the end it will always be about the speaker and room cooperating as one. B&O tried to help achieve that from a speaker point of view which is very clever. No doubt this is a new landmark achievement. I'm hoping it will put DSP on the map of valid options.
 
Last edited:
Discovered this thread a bit late, and not to disgress from the current discussion, but how do the miniature Carver CR Ribbon-sidewoofer speakers fit in here?

I could not find anything related to their technology other than the usual Carver pseudo-techbabble, so how is their design related to a Bessel/Stiles Array?

Eelco