What a beautiful book on Danish audio, and what better way to celebrate Peter L. Jensen’s invention by putting the current best loudspeaker in the world on the cover.
I personally think that when Peter L. Jensen is looking down from heaven and sees the Beolab 90 he’s smiling.

Geoff –

I am glad that they have included an omni mode – if true – I don’t know if there are any drivers on the back of the unit – but omni is still too hot on the direct sound. This can work OK in a very large room, but not in most domestic sized rooms where you are so close to them. Remember, if the speakers themselves are the strongest output in the room then they will be obvious sources themselves and detract from the spaciousness that is possible.

There is no sound field until the recording is played back, and there is no IACC contained in the recording. Both ears are free to hear both (or all) speakers, and crosstalk is a part of the system. It is the speakers and room that create the sound fields in the reproduction, not the recording.

In a nutshell, most speaker manufacturers think that stereo is two speakers beaming their sound into your two ears, where your brain then is “fooled” into making stereo. The room treaters and other experts will even try to damp out all reflections so that you might have the privilege of hearing only that precious direct sound from the speakers. This is a great tragedy and has been going on now for most of the stereo era, to the detriment of all of us.

But NO – stereo is a field-type system in which the function of the loudspeakers is not to be direct radiators but rather to be image model projectors, creating sound fields within the listening room that more closely resemble live sound – a reconstruction of the sound fields contained in the recording rather than some mistaken binaural confusion. What you should hear is an auditory scene spread across the whole front of your room, the speakers disappearing as obvious sources, rather than a spread of sounds between the speakers with little depth or spaciousness. I eagerly await an audition of the 90s, but I have never found speakers that are too hot on the direct sound to be able to do this. I have in my home some speakers that were built in accordance with image model theory principles that I could demonstrate if you are ever in Florida. If the engineers are interested I can describe how to build them and use them.

Gary Eickmeier

Hi Garry,
You claim to have speakers with (according to you) so called live sound?
Because of that (if I may say so) pretentious claim i am curious to hear what speaker brand you are talking about?
What’s the panoramic power response of those speakers?
How closely is the output of a pair of those stereo speakers matched?
What is the dynamic capability/headroom of those speakers?
Are those speakers able to reproduce the 10 octaves of human hearing (20 Hz to 20kHz)?
Did you get a chance to look at some of the links I posted?

Kind regards
Bart

Hi Geoff & Garry,
It wasn’t my intention to come off as rude or to offend anybody.
If I made someone in this post by accident think otherwise I want to apologize for that.

Kind regards
Bart

PS – yes, I have read most of the links that you sent, and Dave Moulton has encouraged me in my research in the past and the speakers he made with B&O use similar principles to mine.

Bart –

Most of what I said didn’t register with you. I’m not talking about 180 degrees radiation in front of the speaker, I am talking about more sound in the reflecting direction – a negative directivity index -6 dB more output toward the front wall of the room. I recall some of Moulton’s writings saying he didn’t want much sound going toward the front wall, just the side walls, which would explain the differences with his Beolab 5 design. But that is not a correct design. Obviously with live sound the centerfill is just as loud and just as important as the sidewall reflections and the general pattern has to be even all across the soundstage. Speaker positioning is also a major part of the theory and I don’t know what Dave has said on that.

When I said that he has supported me I simply meant that he came up to me in 1989 before I presented my paper and gave me some suggestions and some reading, which I had already done and indicated in my references at the end of the paper. Since then I have written to him but he has not been real responsive, not sure why.

Gary Eickmeier

Bart –

Thanks very much for that explanation! I tried to talk to Soren Bech during a meeting at the AES about that project, hoping I could have some input, but never connected up with each other. That was a fantastic opportunity to test all kinds of spatial theories about stereo.

What you have just told me has some bad data leading to wrong conclusions. The reflections I am talking about in Image Model Theory are delayed 10 milliseconds from the front wall and slightly more from the side walls after the corner secondary. After that very important second “hit” of the sound (the first reflections, basically) it does not return in any strength to even be able to build up beyond 50 ms.

Also – and very importantly – those side wall reflections should NOT come directly from the speakers themselves as with a 180° dispersion but as secondaries that were launched from the rear radiation of the speakers. The main reason is to be able to control the Direct to Reflected (D/R) ratio and delay. In fact, having too much direct sidewall reflections might cause the undesirable artifact of the whole soundstage collapsing to one side as you move closer to one speaker. I know this from experimentation. I had my prototype speakers with more output toward the side walls in the front radiation, and that is what happened. Then, when I swapped sides with the speakers and they then had more frontal sound cast toward the other speaker, I had a stable center image as I moved about. See Mark Davis’s work on the Soundfield One speaker at DBX. So the lateral reflections must come from the rear radiation and the front wall of the room must be specularly reflective for this to work. You must position the speakers as I have said and you must have more radiation toward the rear by about 6 dB. The speakers are oriented diamondwise into the room, launching the sound from four driver faces, each one facing 45° from the centerline. The rear faces have full gain, the inner front face has -3 dB, the outer front face is -9 dB compared to the rear faces. Final soundstage will be formed from the real and virtual images of all 8 sources equally, the rear ones being “lit up” by the stronger radiation to the rear.

All of these factors are very important to the final result, none of which would normally occur to the speaker designer or experimenter, which is why I am beating the bushes telling my story. I stumbled upon most of it accidentally while experimenting with my 901 speakers some 35 years ago. My new design has normal drivers and a radiation pattern a little different from those, and conforms exactly to my theories about spatial audio and mimicking a live sound field. Please believe me that not having a reflective front wall is a basic error and to try it my way. You have the absorptive end near the rear so that the sound can’t come back to the front after your 50 ms. But the reflected sound from the front wall gives the depth of image and brings the first reflections around to the sides with a proper D/R ratio and delay.

Gary Eickmeier

Bart –

It would not be possible to relate all that is wrong with that explanation in a few paragraphs here. Let me just point out that the right ear does not hear just the right speaker etc, it hears both speakers, and the same with the left ear. This is not an error with the stereophonic system, it is part of it. It is called crosstalk and it marks the dividing line between stereophonic and binaural systems.

In stereo – a field type system – the object is not to record signals for the ears, but rather record and reproduce sound fields in rooms, just like the live sound fields. There can be any number of microphones, channels, and speakers. Binaural, on the other hand, is a system of recording and reproducing signals for the ears on two and only two channels. It is recorded on a dummy head with ears just like ours and is reproduced so that the channels are introduced to each ear separately with no crosstalk and no room sound, either by means of headphones or crosstalk cancelling speakers in a dead room.

The main clue to how to do stereo is to look at the image model of live sound fields. An image model is a plan view of the sources and all of their first and second reflected sound images as additional sources on the other side of the reflecting walls. The live sound model has all of the instruments all across the soundstage emitting direct and reflected sounds from front and side walls. In order to sound the same, the reproduction image model must do the same. It’s as simple as that in the broad stroke.

Gary Eickmeier

>> the midrange and tweeter are even less omnidirectional so they are sending their sound into a lens that reflects the sound 180 degrees horizontal.
>> The lenses on the beolab 5 where designed to transmit treble and mid-range at a horizontal angle of 180 degrees

Say what ?? Not by my measurements; not, I believe, in Moulton and LaCarrubba’s minds and goals; and not, I also believe I recall, by BeoLab’s own measurements. I am therefore very surprised to read these assertions. While I was not able to measure a BL5 unit outside as I usually do with loudspeakers to analyze horizontal radiation pattern, I did do a close 360deg normalized diff by angle of them in situ (temporally averaged) at Moulton’s studio, and the resultant very nearly equi-omni horizontal radpat curves appeared in SensibleSound magazine 105 (fall 2005) and the Boston Audio Society Speaker newsletter v27n3 the same month. So I believe your description is mistaken.
The BL5 also does not sound at all like a forward-radiating design, even a broad 180deg one.

Serious and uniform refinement of broad (or not broad) wideband beamforming is a extraordinarily important achievement, and I am psyched to audition the BL90. As a longtime owner of Allison and dbx Soundfield designs (as you may or may not know, the former using unique drivers that put out nearly as much sound sideways as forward, the latter the first successful smooth beamforming phased array), I believe I well know how the BL90 will present in its various horizontal radpats. But the precision of your achievement appears to represent a major step forward.

You are wise to lie low :) .

Yes, I saw the polar maps some time ago, which is why I am so psyched to audition and measure the BL90.

Also yes, I cannot fully believe the BL5 measurement shown. Not only does it not seem to correlate with how the BL5 sounds but also it does not seem plausible that 1k (say) directly behind (-180deg) would be down in level so much. A ~1′ wavelength and that size of lensed horn?

Quite aside from its difference (not huge, but significant enough) from my more casual measurement:

https://dl.dropboxusercontent.com/u/5082359/BASS-27-03-0510.pdf

See p23 fig8. 5dB/div, normalized horizontal radiation pattern:

0deg = axis, the flat line; -180deg = the lowest curve, down ~8dB, and usually much less, from say 150Hz to above 10k. There is no reason I can see that the lensed horn would shadow and attenuate the 1k rearward output level quite as BL depicts. If I am reading it right. Mine shows 1k down ~5dB directly behind; yours is quite a bit greater (quieter) than -9dB, is that correct?

Some of this is BL wording and pitch. A design that, broadband, is fully half-loud (-9dB) directly behind (-180deg) is not one I and many other investigators would describe as ‘forward half-omni’.

The production horizontal radpats of Mark Davis’s dbx Soundfield designs (1980s) may interest you guys if you do not know them, phased arrays from back in the day without modern DSP:

https://dl.dropboxusercontent.com/u/5082359/dbx_8199_SF_mfd.PDF ,

pp895 on, his AESJ article.

best, David

Yeah, I am sure environment accounts for a lot, and as I said I was not outdoors the way I normally measure.

But mine are still delta-only, with the mike spatially averaging within the same vertical slice at the same close distance and the BL5 cabinet turned per measurement, … so environmental loading was held constant (and then subtracted). I would therefore expect more similarity, but would have to see the B&O -180deg ones in detail and also averaged a little bit to comment with savvy.

Anyway, interesting. I am not one who trusts hi-rez measurements of loudspeakers generally in any case, hearing-irrelevant as they demonstrably are most of the time (I was invited to give an NYC AES convention panel presentation on this a few years ago). But I would never say that B&O do not know how to measure.

Point taken about beam descriptions, at least when imprecise. What we want is uniformity, following choice of radpat type in the first place. This was known to many well before Toole (and indeed before Davis as a grad student in 1972), although the Toole/Olive blind work of quantification and preference re hor radpat is simply invaluable.

Anyone who said ‘cardioid’ is a prizewinning idiot and ought not to be writing about audio or anything else. Of course, some say that of planar designs, where it’s not really true either, although it’s less false.

My 30yo Stereo Review article was just a partial look at unacknowledged problems w/ 2-way designs (the now widely recognized trades at the crossover among frequency point, driver width and beaming, direct output, and power output, meaning for the listener trades among imaging tightness, honky MT balance, and listening distance), which back then some were saying were inherently superior or something for bogus reasons maybe having to do w/ “phase”. As an Allison owner and longtime dbx engineering colleague of Davis, I knew better, and thought an article was due. At dbx we also were finishing up the voicing of the flagship Soundfield design, comparing with the new Allison IC20 and the Snell A3, which have about as different hor radpats as one could conceive, so I mentioned that.

Do have a vacating vacation.

I have done it only once or twice, when weight and/or weather prevented taking the cabinet outside.

In this case, at Moulton’s studio, no industrial turntable, the mike was handheld perhaps a foot away and with the precision pro dbx RTA1 set to continuous averaging and the speaker playing p/n, I moved the mike slowly up and down vertically along the notional axis, from tweeter height to floor and back, a couple of times, moving to and fro (in and out) hardly at all and sideways left-right perhaps a couple of inches. (With the dbx RTA1’s temporal averaging, none of that matters much; the data are altogether repeatable.) I stored the result in memory. I repeated this identically with the cabinet turned -15, -30, -45, -60, -90, -135, and then -180deg or directly behind. One direction only, symmetry assumed. No variables (interferences) change except for cabinet orientation.

Then I subtract the 0deg axis measurement from all of the others, also from itself to produce a straight line (basline). The resultant splayed set of FR curves by angle is what I published in my BL5 review sent to you.

I have done this only one other time and it was during iterative crossover change voicing with a designer, not for publication.

If I were to do this as a regular method instead of my usual outdoors one (measuring at the same angles but moving the mike up/down, side/side within a cubic foot or two in space at seated ear height 7-8′ away), yes, I would compare the two delta sets, lazy indoors vs laborious outdoors, to ensure that they were very close. Obviously you must diff-normalize in order to view hor radpat changes only.

I should have done that QC check for the BL5 review, actually, for complete method QC hygiene.

bostonaudiosociety.org/pdf/bass/BASS-17-06-9005.pdf , from >25y ago, ago has windy details and examples and extra-windy narrative.