Adam Shulman, consultant at design and consultancy firm SIA Acoustics provides a technical overview of steerable column systems, and speculates on where development of the category may head next.
Beam steering’ or ‘steerable’ loudspeaker products have become relatively common and accepted technology, but the fundamental concepts of their operation are much lesser known and understood.
This type of loudspeaker provides a number of acoustical and logistical benefits over conventional “point-and-shoot” products, or even modern-day mechanically-articulated line arrays. Because steering is accomplished purely electronically, directivity can be more precisely varied to match the audience geometry without any physical change.
Additionally, because the rear lobe is steered in the same way as the front lobe (as opposed to being mirrored across the y-axis, as with a mechanically-steered loudspeaker), lateral reflections can be significantly reduced.
What is ‘steerable’?
Many products purport to be ‘steerable’, but from an objective standpoint, a DSP-steerable loudspeaker must fulfill the following criteria.
Firstly they need wide-dispersion components – ‘steering’ loudspeakers require interaction or ‘overlap’ between components to operate. Ideally, these components radiate hemispherically throughout their entire operating bandwidth, for maximum interaction in the “coupling plane” (the vertical axis, in most cases). Practically speaking, this is generally not the case; conventional transducers ‘beam’ at high frequency, limiting interaction. Some manufacturers use dome tweeters to attempt to combat this, while others hornload components to maintain beam width throughout the pass-band.
A second requirement is ‘small’ component spacing. By ‘small’, we mean compared to wavelength. If the spacing is within 1/2 of a wavelength (at the highest frequency of the driver’s pass-band), energy is focused in one main ‘lobe’. The ability to steer is still maintained even with greater component spacing, but with the addition of deleterious off-axis grating lobes (see figures 1, 2 and 3). The limitation, therefore, becomes one of the higher frequencies.
In order to achieve such small component spacing, output is severely constrained; a physical restriction of their physical size (i.e cone volume) and the material properties and construction currently used in loudspeakers (i.e excursion).
Different manufacturers address this in a number of ways. Some simply ignore this issue altogether and allow their products to exhibit severe comb filtering in the top few octaves. Others manipulate the construction of the transducers used in their products to achieve the smallest spacing physically possible. And others lowpass their product where it begins to perform ‘badly’, since speech intelligibility (the main goal of these products, as they currently stand) is not considered to rely much on this part of the spectrum.
Metrics such as STI (Speech Transmission Index) do not exceed 8.5 kHz, while others such as %ALcons only consider up to 2 kHz. A final requirement is discrete processing and amplification. To achieve even the simplest steering for a range of frequencies, each cell must be capable of independent delay, amplitude control, and band-pass filtering. For more sophisticated steering (for example, to address complex audience geometry), the ability to manipulate frequency- and time-domain processing (via FIR filters, for example) is required.
Today’s products
When it comes to the current marketplace, the ‘steerable’ arena is primarily focused on speech intelligibility applications. As such, it is dominated by relatively low-power, limited-bandwidth products as compared to concert-level technology. The technology to implement steering in this limited manner (DSP cards, small-format amplifiers, drivers, etc.) is readily available and relatively standardised. In large part, modern-day ‘beam-steering loudspeakers’ are characterised by a number of common parameters. Power
Output (SPL) – In most cases, this is restricted to approximately 95-100 dB SPL, typically specified at some distance (several manufacturers give this measurement at 30 meters). This assumes the narrowest directivity possible, focused to achieve the maximum SPL at a specific point in space.
Spectral Bandwidth Capability is actually best divided into two distinct sub-components.
Range of Effective Pattern Control – The spectral range over which the loudspeaker exhibits significant pattern control. This is a primarily a function of the effective line length (at low frequency) and component spacing (at high frequency). Some manufacturers divide the spectrum into multiple pass-bands (i.e. separate high- and low-frequency sections) to improve this.
Range of Effective Output – The spectral range that can be produced by the device. This is primarily the result of the specific transducers’ capabilities. In many cases, it is possible for the loudspeaker to produce sound that is lower in frequency than can effectively be steered a given column length. This parameter is much more commonly cited in specifications, but it is important to distinguish this from effective pattern control.
Effective ‘Steerable’ Range (in degrees) – The polar range over which the column can be steered. This is a function of individual transducer directivity; remember that beam steering relies on interaction between devices. As loudspeakers generally exhibit directivity that is inversely proportional to frequency, interaction (and therefore the ability to vary directivity) will lessen at high frequency. Currently, some loudspeakers are able to steer as much as +/- 70 degrees in the vertical direction, owing to very small, closely-spaced and consistently wide-beam width high-frequency components.
Control and User Interface – The manner in which the user interacts with the technology. There are currently a large spread amongst manufacturers on how this is to be addressed (i.e. via computer, mobile device, on-loudspeaker buttons, etc.) and the degree of user control that should be provided.
Prediction and Processing Implementation Methodology – The means by which the ‘steering coefficients’ (really, filter coefficients) are derived as based on user data entry.
Some manufacturers require direct entry (the user manually determines the quantity, width and angle of ‘beams’). Others request basic information about the venue geometry, and the desired acoustical criteria (i.e. the SPL variation over the audience) and determine a numerically-optimised set of filters. Both have potential drawbacks: the former generally results in a trial-by-error routine and requires user expertise for a good result (though the process is most transparent), but the latter method requires the prediction tool to make a large number of decisions based on limited user input (though the result is truly optimised for the input parameters).
Criteria for the future
The obvious progression of loudspeakers is to harness ‘steerable’ technology on the large scale. We have already discussed the limitations of today’s column loudspeakers. Even advanced ‘modern’ line arrays are an inherently compromised solution, requiring ways of acoustically isolating each element at high frequency to minimise destructive interference (because sufficiently close spacing is not yet possible), while simultaneously maximising constructive interference lower in the spectrum to achieve pattern control, and achieving a smooth transition between the two device types and spectral regions.
Even if this feat is accomplished, there is no great solution with respect to the design of the high frequency waveguide itself. It must be adequate for a range of enclosure splay angles (limiting in and of themselves, as they are not continuously variable) and array curvatures; not too wide to produce comb filtering at the top of the array (where curvature is typically the least), and not too narrow to open up “gaps” in highfrequency coverage at the bottom (where splay is generally the greatest). To attempt to address this, some manufacturers have even gone so far as to produce elements with different vertical patterns to be used throughout the array, but this is only an incremental improvement and imposes significantly increased logistical complexity.
Power output developments are likely to involve improvement in transducer and material technologies, increasing power output per radiating surface area. In the arena of spectral bandwidth capability the lowfrequency limit for effective pattern control will always be a function of line source length. However, more efficient and improvements in wave guide design will permit closer spacing of components, increasing the high-frequency limit to include the entire audible frequency range.
Because components will be smaller, the effective steering range for these products will be larger (perhaps as great +/- 90 degrees, if all transducers can be made to truly radiate hemispherically). This could mean that specialised components such as ‘down-fill’ loudspeakers will no longer be necessary. In terms of control, leveraging the power of mobile devices, on-board DSP and perhaps even mixing console plugins, the need for a discrete computer will be largely eliminated. Access to the loudspeakers will be through universal protocols such as HTTP and languages such as Java. Product-specific software will no longer need to be installed.
As a component of this, steerable loudspeaker arrays will be ‘intelligent’ and ‘self-healing’. On-board sensing circuitry will detect any out-of-tolerance components in real-time and adjust processing or limit output across the entire system to maintain the best performance possible.
Perhaps the most significant issue for the future will be prediction and processing implementation methodology since such advanced technology (far beyond what can be calculated with pencil and paper) is useless without a highly effective control mechanism. With such technical complexity at work, users cannot be required to directly enter acoustical parameters. Instead, prediction software will record the venue geometry.
The user will enter only target performance goals such as absolute SPL, SPL and tonality consistency, and any ‘black-out’ zones where energy should be minimised. The software will convey any trade-offs between these goals (i.e. reduced maximum SPL to achieve improved tonal consistency, or reduced SPL consistency to achieve better isolation of ‘black-out’ zones). Based on this input, a set of complex filter coefficients will be determined and loaded into the loudspeaker, perhaps even using an array’s cumulative on-board DSP to reduce computation time. Any environmental variables (temperature, humidity, nonzero loudspeaker site angle due to rigging constraints, etc.) will be detected by on-board sensors and factored in automatically.
Whilst an exciting and useful technology, beamsteering products have not yet reached maturity and continue to enjoy limited application. Market acceptance and technology continually advance, however. Some may argue that several products available in the marketplace now fulfill these ‘future’ criteria, and this is partially true. Several high-power “steerable” products do exist, but none (to the author’s knowledge) satisfy all of the above criteria; most commonly points falling short with respect to points relating to spectral bandwidth, control, prediction and processing or mounting flexibility.
But the technology embodied in ‘steerable’ loudspeakers is to be the basis for loudspeaker development moving forward. No alternative is as effective; for the acoustician, audience member, engineer, or performer.