B&O Tech: Intuitive Directivity Plots v.2

#48 in a series of articles about the technology behind Bang & Olufsen loudspeakers


In a past article, I tried to come up with an intuitive way of representing the beam width (or “directivity” – if you’re a geek) of the BeoLab 90. I realised after posting, that there is another way to do this which is used in loudspeaker reviews in some magazines (mostly because it’s the way directivity is plotted by MLSSA). So, I’ve taken the same data as before, and re-plotted it using a “waterfall” function in Matlab. It’s just a different way of looking at the same information – but it might be helpful.

If you’re curious about the details regarding the data itself, this is described here.



Figure X: BeoLab 90 directivity vs. frequency as a "normal" three-way loudspeaker with the Beam Control disabled. (Note that this is not possible in a production model.)
Figure 1: BeoLab 90 directivity vs. frequency as a “normal” three-way loudspeaker with the Beam Control disabled. (Note that this is not possible in a production model.)


Figure X: Directivity vs. frequency of BeoLab 90 in Narrow mode
Figure 2: Directivity vs. frequency of BeoLab 90 in Narrow mode



Figure X: Directivity vs. frequency of BeoLab 90 in Wide mode. Compare this to the BeoLab 5 plot.
Figure 3: Directivity vs. frequency of BeoLab 90 in Wide mode. Compare this to the BeoLab 5 plot, below. As can be seen here, the low frequency directivity (or beam width) is controlled to be the same as the mid and high frequency bands.


Figure X: BeoLab 5 directivity vs. frequency.
Figure 4: BeoLab 5 directivity vs. frequency. As can be seen here, the Acoustic Lens ensures a constant directivity (meaning the same beam width at all frequencies) in the high end. However, the low frequency band is omnidirectional.


Figure X: Directivity vs. frequency of BeoLab 90 in Omni mode
Figure 5: Directivity vs. frequency of BeoLab 90 in Omni mode


  1. Millemissen says:

    I must admit, that I found the vs1 ‘better’ for showing what is going on with BL90/directivety. But maybe that is just me!
    By the way – I had no problems with the way you described it originally.

    That said – it is always good to have things explained in different ways. It might help more people to understand how B&O/the acoustic department is working with the technology of loudspeakers.

    Greetings MM

  2. Hi Geoff
    I have since summer followed your blog. I really enjoyed your B&O Tech articles, and impressive to see it materialized in BL90. I look forward too learn about future, smaller speakers with same or similar advanced technology. A speaker that hopefully would fit my relatively small living room.
    Regarding ways of showing what is going on it might be an idea to use visualization, perhaps something like what can be seen on the homepage for the Kii Three speakers?
    Cheers PS

  3. Hi Preben,

    I assume that you’re talking about the animations on the page you mention. These show the concept for one frequency, and assumes that that particular frequency is omnidirectional from the front driver. However, the important aspect of BeoLab 90 is that the beam width (or directivity) is the same at all frequencies. This means that, at some frequencies, where the front driver has a naturally wide directivity, the side and rear drivers are used to cancel the signal off-axis. However, at another frequency, where the front driver has a too-narrow directivity, the side and rear drivers are used to augment the signal off-axis.


  4. Hello Geoff,

    first of all congratulations on your blog which I enjoy weekly and most of all your acoustic work on actively controlled directivity, in my humble opinion after many decades finally a real important development in loudspeakers engineering. I only wonder about the vertical directivity of the Beolab 90 which through floor and ceiling reflexions also pays its role at the hearing position and is usually problematic on non-coincidental drivers, could you post some words or maybe measurements about it?

    Thank you and kind regards,

  5. Hi Theo,


    We may make the vertical directivity plots public in the future – this will depend on a number of factors.

    However, I will say that, due to the “strange” individual magnitude and phase responses of the 18 loudspeaker drivers (required in order to make the Beam Width Control work), a traditional vertical directivity analysis doesn’t work for Beolab 90. Of course, it can be measured – and of course vertical directivity must be considered.

    However, I would say that, in a “traditional” loudspeaker, coincidental drivers does not make vertical directivity non-problematic. It merely changes the behaviour of the lobing. So, by making the drivers coincidental (or just co-axial), you’re just changing the frequency-dependent behaviour of the vertical directivity. You’re not solving the problem, merely changing it.


  6. Hello Geoff,

    many thanks for your kind and quick response, am not sure though with what you say about the coaxial drivers, their vertical directivity behaviour can be almost as smooth as their horizontal one, for example http://www.stereophile.com/content/kef-ls50-anniversary-model-loudspeaker-measurements . Or do you mean that you can’t create constant directivity concepts (especially in the bass) with them? That is of course true, but psychoacoustically less critical then big directivity changes in the mid range.


  7. Hi Theo,

    Having drivers arranged coaxially makes their directivity more radially symmetrical (in other words, the vertical and the horizontal directivities will be more closely matched), but only if the baffle (the front of the enclosure) on which they’re mounted is radially symmetrical. The diffraction caused by the baffle cannot be ignored.

    In addition, the vertical directivity of a non-coaxial system can’t be dismissed simply because it’s not coaxial. As you can see in this posting, the choice of crossover has a huge influence on the directivity in the vertical plane containing the two drivers. Of course, the individual directivities of the two drivers in the crossover region and the baffle also plays a role here. As a simple example, if you are crossing over from a mid-woofer to a woofer at a given frequency where both are individually omnidirectional, and the distance between the drivers is significantly smaller than the wavelengths of the crossover region, and the phase response of the crossover has been managed carefully, taking the responses of the drivers into account, and the baffle is small, then it’s irrelevant whether the drivers are coaxial or not. This could also be scaled to a tweeter / midrange, of course…

    Finally, an issue of coaxial drivers is the modulation of the central driver by its adjacent driver due to the movement of the baffle surrounding it. For example, a tweeter in the middle of a midrange ring will exhibit changes in frequency response with changes in excursion of the midrange. So, you might simplify the issue of vertical lobing in the crossover region, but you may incur a cost at a different frequency range that “infects” all radiation directions.

    In other words: making drivers coaxial is not a panacea. It’s not necessarily a bad idea (there are some good coax loudspeakers out there) – but it’s not necessarily the best solution for all problems… (there are some less-good coax speakers out there – and they’re less-good, in part, because of a bad implementation of a coax design).

    As for your statement that constant directivity (especially in the bass) is “psychoacoustically less critical” than big [vertical (my addition – but implied by the rest of your comment)] direcitivty changes in the midrange – I would firstly disagree, secondly ask for some proof of this statement, and thirdly, ask that you be more specific. Remember that we’re appear to be discussing vertical directivity in the crossover region(s) – not necessarily the “midrange”…


  8. Hello Geoff,

    of course the baffle still plays a role and thus should be taken into consideration as in the previously mentioned KEF or for example the new Genelec 8351A but this of course is also true for non-coax plus the extra problems at crossover frequency.

    And yes, in theory you don’t need a coaxial configuration for that if wavelenghts are smaller, but in the psychoacoustically important transition region (I insist there :-D ) from mid to tweeter in reality drivers are usually too large for that, a reason also almost no real d’ Appolito speakers exist on the market.

    About the issue of modulation, of course its important (as also IMD) that’s why I would always recommend a three way implementation like above Genelec, but not only for coaxial speakers. ;-)

    And of course coax is not a panacea (most actually are imho poor designs except mainly some designs from KEF and Genelec) but we are talking here about how to get the best compromise so we should compare the best coax to the best conventional designs. And in conventional rooms where reflexions from floor and ceiling play a role I have noticed that above Genelec coaxes have almost always a smoother frequency response at the listeners position due to smoother power response over all horizontal and vertical angles then the best conventional monitors with fantastic horizontal directivity but some unavoidable vertical problems.

    Finally as you of course knew my statement about smooth directivity being more important in the mid/high region than lower is purely intuitive due to the psychoacoustic sensitivity in hearing, directional resolution of ear which is much lower at bass, Blauert frequency bands etc, but I could ask you the same way for some proof if you deny it. ;-) Crossover frequency to tweeter is on 95+% of the speakers I would guess between 1,5 and 5 kHz which is what I meant with “midrange”.


  9. Hi Theo,

    It seems to me that, for the most part, we agree. A well-implemented version of something is better than a poorly-implemented, different version of the same thing. (I’d rather have a well-prepared hamburger than a poorly-cooked lobster…)

    Of course, cannot comment about my personal opinions of particular loudspeaker models – especially non-B&O.

    It seems that we disagree about the relative importance of constant directivity in the horizontal plane vs. lobing in the vertical plane – but I would say that this should not be a big enough conflict to stop us from being friends… ;-) Certainly, an optimal situation is where you have both. However, I would add that, in my opinion, it is not vertical-plane lobing in the crossover region that is the issue of concern, but the general vertical directivity of the loudspeaker across the entire midrange-to-high frequencies. In other words, the more omnidirectional the loudspeaker in the vertical plane (inside and outside the crossover region) the more audible the effects of the floor and ceiling reflection on the perceived timbre of the loudspeaker. The results of the ARCHIMEDES project showed this. It may come down to a difference in opinion about the timbral vs. the spatial properties of a loudspeaker pair. Very generally speaking, vertical reflections affect the timbre, whereas horizontal reflections have a greater affect on the spatial representation of the recordings.


  10. Hello Geoff,

    not only for the most part but I even fully agree! :-) with your last post, both about the importance of the implementation, as well of the importance of the general vertical directivity across the entire midrange-to-high frequencies and that horizontal directivity is more important as it additionally also affects the spatial representation.

    Wish you a great weekend,