Article

Thursday, April 12, 2012

Church Sound: Ways To Easily “Ring Out” Your Room To Eliminate Feedback

This article is provided by Behind The Mixer.

 
Ringing out a room is a process for eliminating problem frequencies, wherein those frequencies are prone to feedback in your room. 

You might hear it called “tuning the system” or “tuning the room.”

The last phrase “tuning the room” is actually a wrong phrase to use. You aren’t tuning the room, you are tuning the house EQ to work in the room.

Additionally, tuning the system and ringing out the room are a little different. Ringing out the room focuses on feedback frequencies whereas tuning the system includes dealing with feedback frequencies, but is also aimed at sculpting the house EQ so your system sound is in line with your needs. 

For example, a house EQ for a country music setup will be different than a house EQ setup for heavy metal music.

There are three methods for ringing out the room:

1. ($$$$) Hire a professional. An experienced audio professional can use a real-time audio spectrum analyzer and adjust the house EQ to obtain the desired sound. Not only can they deal with problem frequencies but they can also sculpt the house EQ to be better suited for your needs. Go this route if you can afford it. They can ring out the room but they can also tune your system.

2. ($$$) Use an automated solution. An automated room equalizer like the dbx DriveRackPA can be used to eliminate those problem frequencies for your system. It’s effective but not the best.  Yes, I know dbx claims it a miracle in a rack component, but nothing beats your ears (or the ears of a pro). I’ve known people who have used these types of rack-based solutions for portable systems. In my experience of working on portable systems, I tweak the house EQ during the sound check based on what I hear.

3. (FREE) Ring out the system on your own. The cheapest option and one you can do. You won’t get the sound qualities that a professional would get using method 2, but if you’re dealing with feedback frequency problems, it’s a great place to start.

The eight steps to ringing out the system on your own:

1. Set up your board for a proper master volume level.

2. Set up a microphone on the stage for a person to use for speaking.

3. Have the person talk into the microphone and set up the proper gain structure for their channel.

4. Turn up the master fader until you begin to hear feedback as a slight ringing.

5. Identify the frequency by ear or use a spectrum analyzer. Only use a spec analyzer if you know how to properly use it.

6. Cut that frequency in the multi-band house EQ until it goes away.

7. Repeat steps 4 through 6 until many frequencies tend to feed back at the same time. Then….

8. Reduce the master volume back to the normal level. The person’s voice should still sound natural. If it doesn’t, tweak your house EQ changes appropriately by giving a little boost to those frequency bands you cut.

Going through this process, you’ll eliminate those problem frequencies while also gaining more headroom in your system before experiencing feedback.

Ready to learn and laugh? Chris Huff writes about the world of church audio at Behind The Mixer. He covers everything from audio fundamentals to dealing with musicians. He can even tell you the signs the sound guy is having a mental breakdown.

Registration Opens For AES Conference On Music-Induced Hearing Disorders

The Audio Engineering Society (AES) has confirmed that registration is now open for its 47th International Conference on Music-Induced Hearing Disorders, which will take place at Columbia College in downtown Chicago, June 20-22, 2012.

The conference presents expert knowledge from audio engineers, academic researchers, medical experts and cutting-edge manufacturers, with a total of 18 papers being presented over two days.

“The conference is a great opportunity for people to learn a wide array of perspectives on hearing health in the music industry,” says Michael Santucci, conference chair and president of Sensaphonics Hearing Conservation, one of the conference sponsors. “We have several presenters coming in from Europe, along with experts from several U.S. universities and manufacturers.

“This is a great opportunity for AES members to gain critical knowledge on the issue of hearing health in the music industry, and to network with the leading experts in the field.”

The papers being presented span a wide range of topics relevant to the music industry, including measurement techniques for in-ear monitors and portable music device, new research in measurement and diagnosis of hearing problems, and new hearing health products. In addition to the papers being presented, the conference will also have trade show booths from its platinum sponsors.

Full program details and secure online registration are now available here.

Costs for the AES 47th International Conference on Music Induced Hearing Disorders are $600 for AES members, $700 for non-members, and $300 for students, and includes conference attendance, premium on-site catering (two light breakfasts, two lunches, one full dinner) and related social events.

Hotel room blocks at attractive rates have been reserved for conference attendees at two nearby downtown Chicago hotels. Attendees are encouraged to reserve their spot early to ensure availability.

AES
Sensaphonics

Wednesday, April 11, 2012

Predicting Array Performance: Hanging The PA Right The First Time

Back in the good old days predicting the performance of a group of loudspeakers was a hit and miss proposition. We tried to hit all the people and miss the walls. We were happy if we had enough devices to point a transducer everywhere that needed coverage and enough power to make it good and loud.

Complex interactions between devices operating in the same bandwidth, fine level adjustments for individual devices and precise flying angles were the least of our worries. And even if we were worried about such things, we didn’t have the tools to deal with them.

Then came the digital revolution. With the advent of abundant computer horsepower, remote amplifier control and DSP, our capacity to exercise control over sound system parameters took quantum leaps. At the same time, improvements in test equipment allowed manufacturers to give us meaningful data on the performance of the loudspeakers we were driving.

This data combined with the processing power of the modern personal computer made it possible to actually (GASP!) predict the performance of an array before it was hung. Mark IV Audio (read: Electro-Voice, Klark Teknik, Midas and Altec Lansing) was one of the first companies to bring some of these tools to the masses—the AcoustaCADD program was an early example of sound system modeling software.

They also developed a program called Hang Ten to help Electro-Voice MT-4 owners figure out where to attach flying straps to get the boxes to array properly. And anyone who has herked MT-4s around knows that experimenting with different configurations in the real world just wasn’t that much fun.

Later EV produced a program called ArrayShow, which was extremely useful for demonstrating the summing and cancellation between adjacent cabinets hung or stacked in close proximity.

Bose also had its Modeler software. But these products were manufacturer specific, which limited their usefulness.

The next breakthrough came with the introduction of EASE. Although EASE has a distribution agreement with Renkus-Heinz, its loudspeaker database is an unrestricted club. Anyone can join by testing their loudspeakers in a specific manner and submitting the data in the proper form.

Almost all of the reputable manufacturers have basic cabinet data available on their web sites and product CDs. This allows the system designer to pick and choose different cabinets for different applications, mix manufacturers or even, in the case of some of the big dogs, generate EASE data for their proprietary boxes.

We can use this data to predict coverage and SPL levels in a room, set delay times and volume levels of specific cabinets and even to model complex interactions between devices. EASE also does acoustical predictions including reverb time and intelligibility estimates.

For the audio consultant who has time to painstakingly draw a room and insert all of the appropriate wall and ceiling treatments, this is a great tool. But does this really apply to the touring community?

The line array craze has managed to drag some of us kicking and screaming into the world of predictive software. Line arrays only behave like line arrays are supposed to behave when the cabinets interact properly. The “hang and bang” approach leads to extremely uneven results in the real world. So, almost all of the manufacturers fielding these products have created some software to assist their users.

These are not true modeling packages because, with a few exceptions, they only help you determine vertical splay angles needed to cover angled floors, balconies and the like. The horizontal coverage of most line arrays is a fixed quantity. And level prediction with a line array is frequency dependent in the far field.

The point at which we go from the vaunted 3 dB loss per doubling of distance to typical inverse square law behavior (6 dB loss per doubling of distance) changes with frequency, making broadband SPL predictions difficult. But for most of us using traditional cabinets in traditional clusters, there are some very useful tools out there for making sure we hang what we need to hang and point it where it needs to point.

A program I have employed with success is LARA from Integral Acoustics. It uses EASE data, so most commercially made boxes can be imported. It has a convenient library of pre-constructed rooms that correspond to most of the typical venues we encounter.

They are easily modified to reflect the exact dimensions of a particular space. Defining acoustical treatments is not necessary because LARA treats all surfaces as a perfect absorber. In other words, we are only modeling direct field coverage and not trying to predict how the sound system and the room will interact, so the time required to build a room is reduced to a minimum for the well-prepared house engineer many of the venues have floor plans available on the web.

For those that don’t, getting room dimensions can become part of advancing the show. But even if none of the information is available in advance, this program is so quick that a room can be modeled on site while the truck is being unloaded.

Once the room is built, there are a couple of ways to drop the sound system into the model. There is a library function where you can store all of the cabinet models you are carrying. The speaker locations are chosen by using simple X-Y-Z coordinates. You can build a cluster one cabinet at a time or use pre-designed blocks of speakers that you have defined as clusters.

If you build the array with individual boxes you can change the pitch, roll and azimuth of each box independently. If you build with clusters you must change parameters for the whole unit. Each box or cluster can also be adjusted for output volume and delay.

A chart is generated showing the exact location, height, tilt, volume and delay parameters for each box or cluster. This makes it a snap to put everything where and how it was modeled.

The program generates a color SPL map of the room at chosen frequencies. Audience surfaces can be chosen so that the model only shows coverage where the people are. Or we can look at all surfaces so that sound can be steered away from non-audience areas to minimize the system’s interaction with the room. Individual cabinets or clusters can be turned on or off to see what contribution to the overall coverage is being provided by them.

In addition to SPL maps, the program will provide complex summation maps. These show the constructive and destructive interference patterns that are produced by overlapping coverage patterns. This enables you to see comb filters and the lobing they produce.

These types of tools keep coming along to make our jobs more complicated in some ways but simpler in others. And if we can hang the PA right the first time, we can spend our day tuning, tweaking and sound checking. But the bottom line is better sounding shows with more even coverage through the whole venue.

And that’s what it’s all about, isn’t it? Besides, all this computer modeling will keep us out of the back room on the bus!

Sound Wave Propagation: The Bigger Picture Of What We Hear

Previously (here and here), we’ve been looking at sound on a “microscopic” level, examining particle motion as sound propagates through air.

This time, let’s look at a larger picture of sound wave propagation.

A vibrating object will disturb the surrounding air medium causing localized changes in pressure and particle displacement with the transference of acoustical energy in the form of a sound wave.

Waves can be broadly classified as being either transverse or longitudinal. The distinction for each wave type corresponds to the relationship to which direction the particles in the medium move relative to the direction the wave moves.

For a transverse wave, the particles of the medium move at right angles to the direction of the wave propagation.

Examples of transverse waves include ocean waves, vibrating strings on a musical instrument, and light and other electro-magnetic radiation.

For a longitudinal wave the direction of the wave propagation is parallel (same direction) as the motion of the medium particles.

Sound waves in air are unique in that they propagate as longitudinal waves. Figure 1 shows concepts for transverse and longitudinal wave motion.

Figure 1. Transverse (top) and longitudinal wave motion (bottom) showing transmission medium motion and wave propagation direction. Image credit: Sennheiser. (click to enlarge)

A sound wave in air will propagate away from a source as a wavefront with the speed of 343 m/s. The sound particle vibrations travel outward from the source with the same phase. Sources can take the form of simple geometry, called point or monopole (e.g., single loudspeaker or small hole in a wall), line (e.g., moving vehicular traffic or many closely spaced loudspeakers), plane (e.g., top plate of double bass, large building surfaces, or large two dimensional loudspeaker array), or complex comprising two or more simple sources (e.g., musical instrument or machine).

The wavefront takes on a spherical geometry at a distance from the source that is larger than the source dimensions and are called spherical waves. At farther distances the wavefronts appear to be flat (planar) and are called plane progressive waves. Plane waves are sound waves in the simplest form.

The opposite of a plane progressive wave is the standing wave. And yet another wave type is the cylindrical wave, due to a series of point sources radiating in-phase with each other that results in a line source.

SIMPLE MULTIPLES

Waves, regardless of whether spherical, cylindrical, or plane, can be considered to be simple or complex. A simple wave is a wave that comprises only one frequency, such as a sinusoid. A complex wave comprises a fundamental sinusoid and harmonics, with the harmonics either being simple integer multiples (2, 3, 4…etc.) of the fundamental sinusoid or non-integer multiples (1.35, 2.21, 3.05 etc.), as occurs for many percussion instruments.

Through a mathematical process called Fourier Analysis, we can decompose a complex wave into the fundamental and harmonic frequencies, their relative amplitudes, and phase relationships. Figure 2 shows a Fourier analysis of simple and complex waves.

Figure 2. Fourier analysis of simple (top) and complex waves (middle and bottom). Figures to the left show wave amplitude as function of time. Figures to the right show wave amplitude as a function of frequency (fundamental and harmonics). Top figure is for a tuning fork; middle figure is for a clarinet; and bottom figure for a trumpet. (click to enlarge)

Two or more simple or complex sound waves can combine with each other through an additive process called the law of superposition. The resulting complex wave, assuming that the waves are linearly related, is the sum of the displacements due to each sound source.

Sound waves are linearly related when each is directly proportional to displacement. Non-linear acoustic behavior typically occurs when the source sound pressure level exceeds 140 dB.

While conceptually simple, the law of superposition is complex since the particle displacement (ξ) and particle velocity (u) of each sound wave may not always be in the same direction because the sound waves can arrive from any arbitrary location. Remember too, that particle displacement and particle velocity are functions of time and frequency.

Thus, the law of superposition requires a vector summation of waves. Waves from opposite directions will add momentarily together at a finite point in space and then pass through each other as the wavefronts continue propagating in their respective directions.

Waves can add constructively, resulting in greater amplitude, or destructively, resulting in reduced amplitude. The law of superposition describes this. What determines the resultant amplitude through constructive or destructive addition is the relative phase of the waves.

Waves that are perfectly in-phase (0-degree phase difference) add together with no destructive behavior. Waves that are perfectly out-of-phase (180-degree phase difference) add together to result in effectively zero amplitude. Most complex waves have phase relationships that vary as a function of frequency and do not combine in such simple relationships as described above.

JUST BEAT IT

One interesting wave addition phenomenon is that of beats. Beats occur when two sinusoids of slightly different frequency, typically less than 15 Hz apart, combine at a point in space.

Because the two waves have slightly different frequencies, they will have varying phase relationships, resulting in times when the waves are partially in-phase and partially out-of-phase with each other.

Thus, the waves will add constructively and destructively resulting in slowly varying amplitude.

For example, if the two frequencies are 220 and 229 Hz, the sinusoids will be in-phase and interfere constructively 9 times per second and be out-of-phase 9 times per second, and interfere destructively. The sound level will vary from loud to soft at a rate of 9 Hz.

Most people can perceive beat frequencies up to about 15 Hz. Beyond this value the sensation of “roughness” occurs with no beating. Further separation of the two sinusoids results in perceiving each as a separate frequency. This is one basis for determining the critical bandwidths of the ear.

Figure 3 shows the generation of a beat frequency.

Figure 3. Generation of beat frequency (bottom) from two sinusoids of slightly different frequencies (top and middle). (click to enlarge)

When a sound wave approaches a boundary surface, a portion of the incident energy is reflected and a portion is absorbed by the surface. The absorbed sound is the sum of the dissipated losses within the boundary medium and the portion transmitted through the boundary.

The characteristic impedance of the boundary surface determines the ratio of absorbed sound to incident sound. The physical density of architectural materials is higher than air and results in most of the sound energy being reflected away from the boundary surface.

Two broad classes of sound reflections can occur: standing waves and specular reflections. Standing waves are the result of the law of superposition. Specular reflections are not based on the law of superposition. The sound absorption mechanism described above is applicable to both standing waves and specular reflections.

Standing waves result from interference of two or more waves that repeatedly pass through each other when traveling back and forth between the room boundaries. The result is a wave that appears stationary having regions of maximum amplitude (antinodes) and minimum amplitude (nodes).

ON THE SURFACE

For rooms, the standing waves are referred to as room modes. Three types of room modes occur: axial, tangential, and oblique. Each room mode type is supported by an increasing number of room surface pairs.

Figure 4. Axial standing waves fundamental mode (top), second mode (middle), and third mode (bottom). (click to enlarge)

Axial modes require two opposite room surfaces (one pair); tangential modes require four room surfaces (two pairs); and oblique modes require six room surfaces (three pairs).

Axial modes are the most audible. The tangential and oblique modes are respectively 6 and 12 dB less than the axial modes. Figure 4 shows axial standing waves (room modes).

A specular reflection occurs when the incident angle from the incoming wavefront at the boundary surface equals to the reflected angle from the boundary. This reflection phenomenon only occurs when the wavelength of the incident sound is less than approximately one-fourth the boundary surface dimension.

For the above conditions, the reflections can be approximated as rays and laws of geometrical optics apply. Figure 5 shows simple specular reflection. The wavelength for low frequency sound is often equal to or larger than the room dimensions. When this occurs, there are no specular reflections, and wave acoustics is used for analysis.

Figure 5. Specular reflection where angle of incidence equals angle of reflection. (click to enlarge)

One key concept to remember when sound is incident at a physical boundary is the particle velocity (v) and acoustic pressure (p) are 90° (π/2 radians) out-of-phase with each other. At a boundary, the particle velocity will be zero and the pressure will be a maximum.

This is important when considering sound absorption of materials: the maximum absorption at the lowest frequency of interest will occur at a distance equal to λ/4 from the boundary. At this distance the particle velocity will be a maximum for the frequency corresponding to λ/4. Since most “acoustical” materials are frictional absorbers, a maximum particle velocity will result in the greatest sound absorption.

Resonance is the reinforcement of sound by synchronous vibration. Every object in nature has a “preferred” or “natural” frequency at which the object will vibrate.  Imposing an oscillatory force of the same frequency as the object’s natural frequency will cause the object to vibrate at maximum amplitude will little energy input from the exciting force. Changing the “forcing” frequency by a small amount will effectively decrease the resonant response.

DO THE MATH

When examining a system at resonance, we will observe a maximum peak at the resonant frequency (fO).

The height of the resonant peak will depend on the degree of damping within the vibrating system.

The resonant frequency response can be either very sharp, centered around a high amplitude narrow frequency bandwidth (Δf), or quite broad with lesser amplitude.

The desired acoustical response will determine which characteristic is best.

Systems that have a sharp resonance characteristic are called “high Q”; those with a broad resonance are called “low Q”.

The Q term refers to quality factor and can be calculated by the following equation:

where,

Q = quality factor, unitless
fO = resonance frequency, Hz
Δf = frequency bandwidth, Hz, taken as the -3 dB down points about the resonant frequency

Figure 6 shows both high Q and low Q resonance response.

Figure 6. High Q resonance (without damping) and low Q resonance (with damping). (click to enlarge)

RULES OF THUMB

Try to remember the following, or key them into your PDA or computer “cheat sheet.”

- Most everyday sounds we encounter are complex waves comprising many frequencies.

- Simple point sources radiate sound as spherical wavefronts assuming the wavelength is smaller than the source dimensions. At greater distances from the source the wavefronts flatten out and become planar.

- Sounds combine due to the law of superposition and the resultant amplitude depends on the amplitude, frequency, and relative phases of each wave.

- Waves will reflect from room surfaces. Specular reflection requires the wavelength to be at least one-fourth the room surface dimension.

- Resonance is the response of a system when driven at its natural frequency. The sharpness of the resonance will depend on the damping within the system. Rooms have special resonant phenomena called room modes.

Neil Thompson Shade has 30 years of experience in consulting and teaching acoustics, noise control and sound system design. He is president and principal consultant of Acoustical Design Collaborative, Ltd., located in Baltimore, and he has also been taught acoustics, sound system design, computer modeling and related topics at the Peabody Institute of Johns Hopkins University.

Related articles by Neil Thompson Shade:
Acoustic Fundamentals And The Nature Of Sound
Getting To The Basis Of Everything We Hear

Monday, April 09, 2012

KRK Expands Capabilities & Enhances Integration Of ERGO With Software, Firmware Updates

A new software and firmware update is now available for the KRK Systems ERGO audio recording interface and room correction system designed to measure and analyze phase and frequency problems within a listening environment.

To correct room issues, the internal digital signal processor analyzes and corrects room problems, improving the response at the listening position which results in mixes which translate far better to other playback systems.

This latest version includes support for Apple Mac OSX Lion 10.7.3 and Microsoft Windows 7 SPI (32-/64-bit), and also improves the capabilities of ERGO, such as installation, calibration and configuration.

As part of the update, calibration data can now be stored and recalled, which enables engineers to have multiple settings in one room or to store a set of calibrations for various rooms.

With the new version, users can now also route audio for the 1/4-inch headphone output independently from the master outputs. This allows them to send a customized and dedicated monitoring feed to the musician or singer from any DAW application. This new feature provides a great way for artists to create dedicated cue feeds in their production environments.

Additionally, the ERGO Cal and ERGO Control Panel user interfaces for the system have been upgraded to include all the new functionalities found in this update.

Users with the newest version of ERGO can utilize the system in two main configurations.

In the first, they can use an audio interface to connect with hardware running either Mac or Windows, which allows full access to audio input and outputs for recording and monitoring, as well as room correction.

The system can also be used as a stand-alone room correction device, sitting in-line with a monitoring signal path. Under this setup, users would need to connect to a computer only for the room calibration process.

To achieve the ultimate room correction functionality, ERGO utilizes the RoomPerfect algorithm, licensed from Danish company Lyngdorf Audio, the authority on room correction. This technology uses highly complex test tones, multiple measurements and more than 1,000 EQ points to control an audio environment. The system can derive data on room modes, power response, directivity and LF roll-off.

Licensed exclusively to KRK, ERGO works with all brands of studio monitors. The room correction software comes complete with an ERGO Calibration microphone, FireWire cable, power supply, quick start guide and hybrid installer CD.

KRK Systems

A Road Map Of Future Loudspeaker Optimization

An understanding of the sound reinforcement system optimization process as described in my previous article (here) leads to what may be a startling conclusion: with relatively little effort, this technology could be leveraged immediately.

A number of products are currently on the market that begin to leverage this technology. These range from stand-alone beam-steering column loudspeakers from companies such as Renkus-Heinz, Tannoy and EAW to higher-power arrayable products such as Martin Audio MLA, Duran Audio Target, and Renkuz-Heinz IC Squared.

On the most basic level, the requirements for an optimized system include:

An accurate loudspeaker model. Nearly every legitimate loudspeaker manufacturer has the means of generating basic complex (in the mathematical sense) polar data. The question of accuracy arises when one looks in detail at the mid- and low-frequency parts of the spectrum, and the effect of boundaries (either in the measurement environment, or due to adjacent loudspeakers in an array).

The virtue of various measurement and calculation techniques is not within the scope of this article but generally speaking, the more accurate the model is, the closer the predicted result will be to reality.

A direct interface between the optimization program and the sound system DSP. The results of the optimization algorithm will at least initially include both mechanical (splay angle, etc.) and electronic parameters.

To successfully implement the latter, the optimization program should have direct access to the loudspeaker digital signal processing and amplification on a per-transducer basis. That is, the best performance is gained when each transducer (or at least transducer type or pass-band) can be adjusted independently.

Adding Precision

What type of systems can this apply to? In practical terms, any that satisfies these criteria! In the case of existing arrayed loudspeakers (line or otherwise), the manufacturer prediction tool, in addition to calculating optimized splay angles, would also compute different DSP parameters for each amplifier channel and download these directly into the system processing.

The possibilities for improvement will depend on the resolution of the processing (how many components are on each channel), processing power available (number of FIR filter taps), and accuracy of the loudspeaker model. Assuming that these logistical challenges are overcome, the performance advantages could be substantial, while requiring little more than a software update from the user. Upgrades could be strategically marketed, breathing new life into even aging systems (“Presenting model XX loudspeaker, now with optimization technology!”).

Frequency responses sampled on audience for uniformly driven array. Courtesy Martin Audio (click to enlarge)

Because manually gain- and equalization-shading arrays is a common practice and acknowledged to improve consistency through the audience, it is not unreasonable to expect that much more precise implementation of these techniques would produce far better results, even without any physical component change. Gone would be the days of turning down the loudspeakers at the bottom of an array, or boosting the HF for those at the top!

Some manufacturers provide a rudimentary version of this, whereby “array compensation” (i.e. mid- and low-frequency reduction to compensate for coupling) or ‘atmospheric’ correction (i.e. high-frequency boost to compensate for air attenuation) is provided based on some predetermined functions and assumptions. However, this generally does not factor in mechanical articulation or the audience area geometry. Though not a trivial effort to implement, the jump between this functionality and a truly ‘optimized’ system is not so far.

More Capabilities & Control

Once the pro audio industry has become comfortable with the basic practice and implementation of optimization, the capabilities could be readily expanded.

As CPU power continues to improve, optimization time will decrease as performance and precision increases. Ultimately, processing time will become so short that these systems will be able to adapt to changing site conditions in real time.

As an example: utilizing self-monitoring and the performance criteria entered by the user, a loudspeaker array could automatically recalculate processing parameters mid-concert to maintain reasonable performance upon detection of a driver failure or environmental change such as temperature or humidity increase.

Similarly, by monitoring each component with an understanding of its role in the system, advanced limiting functions could ensure that the system maintains coverage and tonality, even if some driver is operating at capacity. Anyone who has had a particular pass-band limit before the rest will understand the impact that this phenomenon can have on the performance of a system.

Because we have made the shift from defining performance at the loudspeaker to defining performance at the audience, the operator should also have the convenience of addressing location-specific issues by the actual audience area instead of the part of the array responsible for covering it.

To account for acoustical inconsistencies throughout the space, the operator could make equalization adjustments in specific parts of the audience, or ‘zones’ in software – but in this case, these are audience zones and not loudspeaker zones. The user need only define what they want to achieve in the actual audience area, and the optimization program (and loudspeaker DSP) determines what is necessary to achieve that.

As we’ve said, the operator should not be concerned with what is electronically or physically required to achieve that additional cut at 2.5 kHz in the upper balcony. At low frequency, where it may be more difficult to isolate specific parts of the venue, it is the duty of the optimization program to alert the user as to the available ‘spatial resolution’, or the effect of what they are doing in one area on the others.

Fast Forward

An ultimate goal for “optimized” systems of any scale should be decreasing or eliminating dependence on mechanical adjustments. This means that arrayed loudspeakers will require no articulation or splay, perhaps with the exception of extreme down-fill (which could be, as they are today, a different design that incorporates some pre-determined curvature).

Frequency responses sampled on audience for profile (2 dB – mix – 2 dB) optimized array. Courtesy Martin Audio (click to enlarge)

As noted, though steady progress is being made in this area by way of “steerable” column loudspeakers, it is yet to be fully implemented in large or even medium-scale systems. But if the optimization criteria and process are accepted and well implemented, it is not a big jump for the user and in fact even simplifies the situation as compared to optimized systems with both electronic and mechanical variables.

When loading in for a concert, for example, the instruction to the crew changes from “set the splay angles and cable the arrays as per this spreadsheet” to “hang X boxes there and plug them in.” With a system that self-recognizes quantity, position, role, and relationship to other components, no further details are necessary.

Since everything is handled in software, this technology can also easily be expanded to other configurations, such as distributed systems. In this case, delay times, band-pass filtering, and the directivity of each distributed point could be calculated automatically.

Lest some live sound engineers read these articles and assume that this technology is destined for installed systems only because of time or complexity, think again. With “smart phone” technology, the user will be asked for progressively less raw data, leveraging features such as wireless communication, GPS, on-board cameras, inclinometers and accelerometers to gather all of the information needed about the venue.

And with increasing prevalence of digital consoles capable of running third-party plug-ins and integrating into networks, the operator won’t even need an external computer or processing box to set their system up. What may now appear as somewhat of a science experiment will become common, just as it did for any of the digital devices which are now commonly used: consoles, audio networks, amplifiers, etc.

So as to not make it too easy for the manufacturers, we will discontinue our discussion here. But suffice it to say, this article has only just scratched the surface of what’s to come for this area of our industry!

Adam Shulman is a consultant with SIA Acoustics, a leading acoustical and system design firm with headquarters in New York City.

Friday, April 06, 2012

Graphing In 3-D: Another Parameter To Account For Space

Two-dimensional graphs are useful for displaying sound system component specifications, but a third dimension is required for considering parameters that are a function of (depend on) a position in space.

This includes two data types that are absolutely essential to the sound system designer - loudspeaker directivity data and computer models of auditoriums.

Let’s investigate the use of three-dimensional graphs for describing the radiation from loudspeakers.

Why is this data needed? Sound radiation is a three-dimensional disturbance that (in most cases) expands as it propagates. The wavefront from a very small source might resemble the shape of a spherical balloon as it is inflated. It’s convenient to consider that sound energy radiation has as its origin a single point in space (a “point” source).

While not physically possible, this theoretical source forms the basis for understanding how energy radiates from all types of realizable sources. The radiation balloon from a point source would expand uniformly in all directions as it propagated outward from the origin. This source is referred to as a monopole, and the radiation pattern as omni-directional.

If all loudspeakers were omni-directional, there would be no need to describe their radiation graphically. One could use the simple inverse-square law to determine the attenuation of the pressure wave as it propagates outward from the source (sound doesn’t die out, it spreads out).

Conveniently, the attenuation would be the same in any direction. It can be shown geometrically that the area of an expanding sphere grows with the square of the increase in radius - double the radius and you get four times the surface area (Figure 1).

Because the coverage area expands, the energy density collapses, and the result is a 6 dB drop in sound level for each doubling of distance (quadrupling of area) from the origin. Note that this is only true for a point source. Other source types (i.e. line or planar) do not follow the inverse-square law.

Figure 1: A point source energy radiator will produce a wave front that attenuates by 6 dB each time the distance from the source is doubled. (click to enlarge)

A point-source radiator would be of limited usefulness in a sound system. Most loudspeakers utilize interference to concentrate the sound energy in a specific direction. Pattern control horns, room boundaries, and even cupped hands are means of controlling sound radiation.

Like camera lenses of differing focal lengths, a wide variety of patterns are necessary and available. A graphical means of representing the patterns is useful to allow comparisons between various devices.

POLE POSITION
The oldest method of graphing directivity is the two-dimensional polar plot. The loudspeaker is measured at an appropriate angular resolution and distance and the relative levels plotted on a polar graph. Both horizontal and vertical plots are measured.

The coverage angle of the loudspeaker is defined by the -6 dB points on the plot relative to on-axis. A popular resolution is 1/1-octave at 10-deg rotation (Figure 2). While better than nothing, the polar plot has been almost completely supplanted by the three-dimensional attenuation balloon. But first, some background.

A circle can be divided into 360 degrees. If one is mounted on “gimbals” and rotated 180-degrees, the surface area of an entire sphere is swept. So, the coverage angles that describe spherical radiation (omni-directivity) are 360 degrees by 180 degrees (Figure 3).

This fact makes a great trivia question for parties and after-dinner conversations. If one measurement were performed per degree of rotation, nearly 65,000 measurements would be required to characterize the spherical radiation from a single loudspeaker!

Figure 2 (at left): The coverage angles of a loudspeaker are defined by the -6 dB points on the radiation ellipse relative to on-axis. Figure 3: The coverage angles that describe spherical radiation are 360 by 180 degrees. (click to enlarge)

A more practical angular rotation is 10 degrees per step, with the next meaningful resolution increase requiring a halving of the rotation angle. This makes 5 degrees the “next better” resolution, then 2.5 degrees, and so on.

Because each halving of the rotation angle quadruples the number of measurement positions, we can’t get too carried away. Five-degree angular resolution (Figure 4) requires over 2,500 individual measurements per loudspeaker! This represents the practical limit of current hardware and software, and approaches the point-of-diminishing-returns on the information density.

ATTENUATION BALLOON

An attenuation balloon is measured by placing the loudspeaker onto a “positioning” apparatus that allows the loudspeaker to be rotated around an origin and aimed in any direction (Figure 5). A precision microphone is used to measure the sound level at an appropriate reference distance in the far field of the loudspeaker. (That’s a different topic - here we simplify it to mean “at a remote distance.”)

The highest sound level (usually on-axis) is used as a reference and the levels at other angles are plotted in relation to it. The resultant “balloon” will reveal the loudspeaker’s three-dimensional directivity characteristics as a function of frequency (Figure 6).

Two important parameters can be determined from this plot - where sound is going, and where sound is not going. The interaction of two or more loudspeakers can be determined by “superposing” their balloons. This is the heart and soul of loudspeaker array design, as it allows the relative level and arrival time of sound from the array to be determined at any point in three-dimensional space.

Figure 4 (at left): A practical measurement resolution is 5 degrees rotation on 1/3-octave frequency centers. Figure 5 (top right): A positioning device is used to rotate the loudspeaker about its origin at the desired angular resolution. Figure 6 (bottom right): An attenuation balloon displays the sound radiation of the loudspeaker relative to the on-axis radiation.  (click to enlarge)

If high enough angular resolution is used, most of the energy leaving the source has been documented. Now some useful ratios can be derived from the data. The ratio of the on-axis squared sound pressure to the average squared sound pressure is the loudspeaker’s axial directivity factor, or Q.

Expressed in decibels, the Q becomes the directivity index, DI. Q and DI were considered as a function of frequency on a two-dimensional graph in last month’s article. So, now we can have a 2-D graph of directivity versus frequency for each point on the sphere.

That’s a lot of graphs, so only the axial position is usually plotted in this manner. Loudspeakers suitable for sound reinforcement applications have directivity factors >1, often by one or two orders of magnitude (x 10, x 100).

FINISHED PRODUCT
In Figures 7 - 10, we see some frequency-dependent radiation balloons along with their two-dimensional polar plots. The angular resolution is 10 degrees, and the frequency resolution is one octave. The smaller balloons at high frequency indicate a tighter radiation pattern (higher Q).

At low frequency, most loudspeakers radiate energy omni-directionally (Q = 1). Both extremes have been the bane of loudspeaker designers since the invention of the device. Successful designs attempt to increase Q at low frequency where sound wants to go everywhere and reduce it at high frequency where it wants to focus.

Figure 7: 500 Hz attenuation balloon. (click to enlarge)

Figure 8: 1 kHz attenuation balloon. (click to enlarge)

Figure 9: 2 kHz attenuation balloon. (click to enlarge)

Figure 10: 4 kHz attenuation balloon. (click to enlarge)

Pat and Brenda Brown own and operate SynAudCon, the leading independing professional audio education source, with training sessions held around the world and online. For more info go to www.synaudcon.com.

Thursday, April 05, 2012

Science Or Snake Oil? The Facts Behind The Hype About Loudspeaker Wire

Too many good folks have been separated from their hard earned money by hyperbolic claims about loudspeaker wire. There will always be people with more dollars than sense, but they don’t last very long in professional audio.

I speculate there aren’t many (if any) of you who would pay thousands, or even tens of dollars per foot for speaker wire.

A very basic practice in merchandising is called differentiation. Marketers must come up with reasons for why you should buy their wire. To claim that their wire is better, they must first identify, in some cases invent, a difference.

This search for a selling proposition has sometimes focused on “skin effect.” It’s a real effect and describes how at very high frequencies, electrons travel in the outer layer or “skin” of signal conductors.

Another related property is that high frequency signals travel faster than low frequencies through the same cable.

These phenomena are dealt with appropriately in very high frequency applications with several techniques. For example, one particular type of botique wire is made up of a large number of very small conductors braided or woven into one cable, producing a large surface area or “skin” for a given cross sectional area.

Another approach for high power high frequency power transfer is to use a hollow conductor, resembling a section of copper tubing. If the electrons are going to ignore the center of the conductor, why pay for it?

This is not an issue for audio professionals, working at mere audio frequencies of 20 Hz to 20 kHz. Perhaps it would be if we were sending audio over many miles, like the telephone company in its pre-digital days. They had to periodically correct for waveform smear.

But at the speed that electricity travels, our typical path distances are much too short to be an issue.

Out Of Perspective
Wire is not very sexy or easy to create real marketing hooks for, but it can actually make an audible difference. The dominant mechanism is simple resistance.

It’s perhaps ironic that the “snake oil” markers of loudspeaker wire will exaggerate some real but insignificant parameter far out of perspective while compromising the real deal.

Forget the hype, what’s important for loudspeaker wire is that it exhibit low impedance that is resistive in nature. If the wire has a significant impedance component (reactance) that changes over the audio frequency spectrum, this can form a simple divider with the loudspeaker’s resistive impedance and cause a frequency response error.

In addition, since loudspeaker impedance will vary quite a bit over frequency, even a perfectly resistive speaker wire will cause errors. The magnitude of this frequency response error will increase proportionately as the wire’s resistance increases.

Purveyors of “funny wire” don’t bother to make claims about useful metrics like resistance since that is already defined by the wire size or gauge (known as “American Wire Gauge” or AWG for short). That would be like advertising how many quarts were in their gallons!

However, frequency response errors caused by wire resistance are one of the very real things that people actually do hear. I find this following anecdote instructive.

From a discussion with one individual who was certain that he heard a significant improvement when using his “Snake-O Special” loudspeaker wire (name changed because I don’t remember it), I determined that the wire gauge he was using was marginal for the length of his run. The wideband loss of volume caused by a wire’s resistance will be very difficult to hear without a side-by-side comparison.

But the difference in amount of loss caused by the loudspeaker’s changing impedance at different frequencies can easily cause a frequency response error that is probably what he heard. It’s easy to imagine how a rising impedance at high frequency could cause a pleasant sounding treble boost. Just listen to how clean and clear these “Snake-O Specials” sound!

There are several strategies to manage these real losses from wire resistance. The obvious one is to throw more copper at the problem. Heavier gauge wire with lower resistance will exhibit lower losses for a given run length.

Another fairly obvious approach is to locate the amplifiers as close as possible to the loudspeakers to keep the run length as short as possible. A third less obvious approach is to scale up the intermediate signal voltages.

Constant Voltage
There are cases, such as in large distributed sound systems where neither of the first two approaches is cost effective.

You can’t afford to put a separate amplifier at every loudspeaker location, and sending sound sources over long distances with acceptable losses would require very heavy gauge wire.

The solution borrows a strategy from high voltage power distribution systems such as the one used by utilities to bring electrical power to our homes.

The power developed within a given load increases with the square of the terminal voltage (E^2/R). However, wire’s losses only increase linearly with current flow, because the voltage developed across the wire is a simple function of its resistance times that current. 

Power engineers determined that by raising the voltage carried by transmission lines they could increase the power being carried exponentially while simultaneously reducing the losses due to current flow.

The utility company accomplishes this magic with step-up/step-down transformers. By “transforming” a typical 100-amp at 240-volts residential service, up to tens of thousands of volts at the transmission line the 100-amp draw is reduced to the far more manageable level of 1 amp or so. Wire losses are 1 percent of what they would otherwise be.

Similar manipulations go on in “constant voltage” distributed sound systems but rather than stepping up the voltage to thousands of volts the standard for U.S. systems is 70-volt, with Europe using a slightly higher 100-volt standard. The rest of the world tries to conform to one of those two standards.

Of course, the audio signal isn’t actually held constant. The voltage at rated power is. Both 5 watts and 500 watts constant voltage systems deliver the same nominal voltage for distribution.

The goal in any effective distribution system is to deliver as much power as possible to do useful work in the load and waste as little as possible heating up the wire. In a simple distributed sound system sending a few watts of announcements across a few hundred feet of factory floor, the typical low voltage system could drop as much power in the speaker wire as would reach the loudspeakers.

By stepping up to 70 volts and back down again at each loudspeaker the balance of power delivered versus lost is more respectable.

To put numbers to this concept, say we are trying to deliver 1 watt each to two loudspeakers located 100 feet distant from an amplifier using 24 AWG wire. Because we must count wire losses from the feed coming and going, 200 feet total of 24 AWG exhibits resistance of approximately 5 ohms.

Figure 1: Two different ways of realizing one watt at two loudspeakers. Click to enlarge.

To realize 1 watt at each loudspeaker, there would need to be more than 4 watts into the wire at the amplifier end. (Over 2 watts gets wasted as heat in the wire).

 
If we first step up the audio to a nominal 70-volt level the current drops to such a low level that the same wire would only waste 0.14 watts while delivering the same 1 watt each to the two loudspeakers.

As useful as constant (high) voltage systems are for managing wire losses, they don’t make sense for point-to-point runs in sound reinforcement systems. The main drawback is the size of the step-up and step-down transformers required.

To put this in perspective, the size of the transformer has to double every time you drop the frequency an octave. To cleanly pass 20 Hz both step-up and step-down audio transformers would have to be three times the size of a conventional amplifier’s 60 Hz power supply transformer.

Keep It Short
The good news for most live sound applications is that we don’t have to tolerate extremely long wire runs. By locating power amplifiers near the loudspeakers we can keep wire runs reasonably short. At these shorter distances we can easily afford heavier gauge wire.

While power losses are now manageable, it is worthwhile investigating the next dominant consideration in sizing loudspeaker wire.

Frequency response errors will be caused by the voltage divider created between the wire’s fixed resistance and the loudspeakers changing impedance versus frequency.

Figure 2 and Figure 3 show two representative loudspeaker impedance plots, pulled from the Internet.

These are not offered as either worst case or typical.

From the impedance plot in Figure 2, if we ignore the extreme low frequency, this loudspeaker exhibits a maximum impedance greater than 17 ohms, with a significant region of the upper bass down around five ohms.

Figure 2: This loudspeaker exhibits a maximum impedance greater than 17 ohms. Click to enlarge.

Meanwhile, Figure 3, while more complex, covers a similar impedance range, with a maximum around 16 ohms and a minimum around six ohms.

 
To derive a frequency response error we need to compare the drop at maximum impedance to the drop at minimum impedance. The equations below calculate that drop for a given wire resistance.

Note: To simplify this analysis we will assume all loudspeaker impedances to be resistive. While not strictly accurate, loudspeaker impedances will typically be resistive at impedance minimums and any errors caused by load phase angle at the impedance maximums will not be significant for the sake of this analysis.

Minimum Voltage drop= V max = Z max /(Z max +Z wire) 
Maximum Voltage drop= V min = Z min /(Z min + Z wire)

Frequency Response deviation= FR max = -20 Log10 (V min/ V max)

Solving for 1-, 0.5-, and 0.1-ohm wire resistance we get:

Loudspeaker….......1 ohm…...... 0.5 ohm….. 0.1 ohm

Spkr 2 (16/6)....... -.81 dB…..... -.42 dB…... -.09 dB

Figure 3: While more complex than the loudspeaker in Figure 1, this covers a similar impedance range, with a maximum around 16 ohms. Click to enlarge.

Another related consequence is how wire resistance degrades effective damping factor.

While damping factor is usually though of as a power amplifier characteristic, in reality the wire selection can easily dominate actual damping available at the loudspeaker.

In the above examples the 1-ohm wire would by itself cause a rather weak damping factor of 5 or 6 (regardless of the amplifier’s rated damping factor).

Using the 0.1-ohm wire predicts a more respectable 50-60 damping factor, with some small additional degradation due to the amplifier’s output impedance.

Damping factor deserves a more extensive discussion, but for this exercise we will assume that the amplifier’s output impedance is small with respect to our wire’s resistance.

Gauging Gauge
It’s difficult to predict a precise threshold for audibility of frequency response errors.

Controlled listening tests have suggested that differences as small as a few tenths of a dB can be audible.

To satisfy the dual goals of minimizing frequency response errors and not degrading damping factor for the example loudspeakers selected, I am comfortable with targeting a total wire resistance on the order of 0.1 ohm.

Wire’s resistance varies linearly with length. To keep the total resistance below our target limit of 0.1 ohm we must first project the length of our desired wire run, and then select a wire gauge whose resistance per unit length keeps us within the total resistance budget.

Don’t overlook that the wire length is actually twice the run distance as we must consider the feed to and return from the loudspeaker as effectively in series. We must also add in contact resistance for the connections at all ends.

Lets look at how this works out for a practical example of a 20-foot run. First, we double that to 40 feet to establish the true signal path length.

Then we need to account for contact resistance. I’ve seen Neutrik Speakon (or copies of that connector) rated as low as 1mOhm (1/1000th ohm) per contact when new, and guaranteed

< 2 mOhm over life.

Because there are four connections in our total path lets budget .008 ohms for connections. Subtracting this 0.008 ohms from our 0.1-ohm target leaves us .092 ohms for wire. Dividing this 0.092 ohms by the 40-foot length calculates out to 0.0023 ohms per foot.

Plugging this into the equation for wire gauge:

AWG = 10 ×log 10 R +10 (note R is per 1000 feet)

We get:  AWG = 10x log 10 (2.3) +10 = 13.6 gauge

This is a little cumbersome, but once you have established an appropriate gauge for a nominal run length with your specific system. This gauge can be scaled up or down for other run lengths.

Wire resistance changes linearly with length. It changes non-linearly with gauge. A convenient property of wire gauge is that the wire’s resistance will double for every 3-step increase in gauge (AWG). Conversely the resistance will drop in half for a three-step decrease in gauge.

Based on this same example and rounding off to 14 AWG, we can expect similar performance from a 40-foot run using 11 AWG wire, and a 10-foot run would only need 17 AWG. This numbering convention gets a little unusual below “0” AWG.

One step below (larger than) “0” is “00”, and “000” is two steps larger than “0”. I don’t expect to see speaker wires this large, as they would be very difficult to effectively interface with amplifiers and loudspeakers.

Using this example to size wire for your system will get you in the ball park, but it will be more accurate to use actual impedance specifications for your loudspeakers. Manufacturers of professional loudspeakers routinely publish this information.

Remember, use only the impedance max/min deviation within the audio bandwidth of interest. It doesn’t matter what a tweeter’s DC resistance is or a woofer’s 20 kHz impedance, since you won’t be listening to them there.

You also may want to tighten or relax the acceptable frequency response deviation. Better yet, look at your loudspeaker’s typical frequency response and determine if the response errors caused by your wire losses are additive or corrective.

While I don’t suggest trying to dial in corrective equalization using wire losses, if the error is making your system flatter you can afford to be less aggressive in sizing your wire AWG as long as you keep damping and power losses under control.

John Roberts is a long-time professional audio product and system designer and has been writing outstanding technical articles over the past two decades.

Tuesday, April 03, 2012

Rational Acoustics Smaart v7 Integreated In New Apex Intelli-Ware Software Control Platform

Rational Acoustics Smaart v.7 measurement data is now integrated in the new Apex Intelli-Ware software control platform for the Intelli-X2 and the Intelli-Z² system management processors.

Intelli-Ware 1.4 offers integration of Smaart v.7 data via the Smaart v.7 API (application program interface) enabling transfer of measurement data from Smaart 7 to Intelli-Ware.

Smaart’s transfer function and/or spectrum analyzer response can be seen as an underlay in all parametric EQ bode-plot and crossover windows.

Virtually all of Smaart’s measurement control features including averaging, banding, delay location and measurement selection can be accessed directly from within Intelli-Ware, removing the need for switching from one application window to another when tuning a system. Smaart can run on the same PC as Intelli-Ware or on another PC on the same network.

“Apex has done an incredible job of integrating Smaart v.7 measurement data into their Intelli-Ware control software.” says Jamie Anderson, Rational Acoustics CEO.  “The interface is fast, intuitive and beautifully designed. This is a powerful tool and a great time-saver for Intelli-series users.”

“In line with the continual development of Apex’s Intelli-series of system management processors (including the IX² and IZ² devices), we are very excited about the Smaart v7 integration in the new Intelli-Ware 1.4.” says Jeroen Sierjacobs, marketing manager of APEX. “Intelli-Ware is a powerful control application but it is very easy to use.”

The Smaart v.7 API software development kit is distributed free-of-charge to manufacturers seeking the ability to integrate Smaart v.7 data into their own control interfaces.

Intelli-Ware is available to download free of charge at www.apex-audio.eu

Rational Acoustics

Friday, March 30, 2012

Understanding Specification Sheets: What Do The Charts & Graphs Really Mean?

For the majority of humans, there is nothing simpler than listening to sound. You simply, well, listen.

When it becomes necessary to describe the listening experience analytically, however, a host of complex equations and diagrams are required to describe even the simplest of sonic events.

The benefit of mathematical analysis is that it can yield insights that are not apparent through intuition alone.

Acoustic signals are easily measured, and the audio components that produce them have characteristics that can be measured.

We do not expect specifications to tell us how a product sounds. This is what listening is for.

The main purpose of specifications is to allow us to make sure that we have the right tool for the job, and this information is most often presented in the form of charts and graphs.

But what does this information really mean?

Variables
The heart of understanding the specification sheets that describe audio products is the understanding of dependent and independent variables.

The concept is one that most people use every day, though often without realization.

An independent variable is one that describes a series that has a fixed value.

For example, the time of day in the city that you live in is an independent variable. Regardless of what happens tomorrow, time will progress like it did today.

What will change are your moment-to-moment activities. These events represent a dependent variable. They depend on time.

If you look at a page in your day planner, you are looking at a plot of activities vs. time.

Time is the independent variable. It is the same on every page of the planner.

The scheduled events are the dependent variables, because where you go and what you do depends on what time it is. Most graphs show the relationship between dependent and independent variables.

Now let’s look at a variation on the theme. Let time be the independent variable (it usually is) and let the loudness of the sound system during a show be the dependent variable.

The plot might look something like Figure 1.

Figure 1: In this example, time is the independent variable while loudness is the dependent variable (click to enlarge)

The horizontal axis represents time (the independent variable) and the vertical axis represents loudness (the dependent variable).

We will call the horizontal axis the x-axis and the vertical axis the y-axis, although any two letters would do.

The values on each axis are usually discrete, meaning that they are individual samples, points, or measured values called data points.

The fact that most graphs look like squiggly lines just means that after many data points were taken, they were joined with a line to make it easier to read.

Such two-dimensional plots are found on virtually every good specification sheet in existence. They simply answer the question “What is the value of y when the value of x is this?” Some examples of two-dimensional plots found in audio engineering include:

Y-Axis————————————————X-Axis
Amplitude—————————————- Frequency
Impedance————————————- Frequency
Directivity—————————————- Frequency
Phase———————————————- Frequency
Amplitude—————————————Time
Level————————————————Time

Each plot shows the value of y for a given value of x. Pretty cool. In math-speak, in each case it can be said that y is a function of x. (We sound smarter when we say it like this.)

From this example, it can be seen that frequency is a very common independent variable in the world of audio and acoustics. The y parameters are said to be frequency-dependent.

In audio and acoustics, almost all parameters that we care to know anything about are frequency-dependent. This means that the answer to virtually any question regarding any of the y parameters is “it depends.” Y depends on x.

An example of a frequency-dependent parameter is the setting of a graphic equalizer. In fact, it’s a really good example because it is basically an xy plot of the type that we have been describing.

The x variable is frequency, and the y variable is relative level. The y value depends on the x value.

When you look at the front panel of a graphic equalizer, you are looking at an xy graph, which is why it’s called a graphic equalizer.

What Time Is It?
Another common independent variable is time. Many parameters in audio and acoustics are time-dependent. Examples include loudness, temperature and background noise, just to name a few.

Note that Figure 1 just gives us values. It’s still up to us to know what they mean and how to apply them.

Graphs are valuable because they give us some visual feedback regarding trends in the data. For instance, a glance at Figure 3 (later in this article) shows that the loudspeaker’s on-axis directivity is increasing as a function of frequency.

This means that everyone in the room might hear the low-frequency events, like a bass guitar, but only those in front of the loudspeaker will hear the high-frequency events, like the crash of a cymbal.

It’s clear why we would want the directivity of a sound reinforcement loudspeaker to be “frequency-independent.” The directivity of such a device would be a straight horizontal line.

It’s also important to consider the resolution of the graphed data. The closer together we place the points on the x-axis, the less likely it will be that we missed a significant data point when we measured.

For example, we could take the page of a day planner and break the time axis down into hours, minutes, seconds, or even fractions of a second.

Obviously, there is a point of diminishing return on resolution. It must always be appropriate for the data being plotted.

If you were plotting the arrival time of the tweeter in the main array to the back of the balcony, then one millisecond resolution would be meaningful.

But that same resolution would be extreme overkill for plotting your daily schedule.

What time resolution do I need? Again, it depends!

Following are some examples of common plots found on data sheets, with plain English descriptions of what each one means.

After digesting each, download some data sheets from various manufacturers and attempt to interpret them.

Use them to form an understanding of the product, what it does, and how it might compare to a similar product.

Remember that to fully describe the performance of a product, and infinite number of graphs would be required.

Most “one-number” ratings in audio and acoustics have little meaning.

They usually over-simplify something that is much too complex to specify with a single number.

Unfortunately, many people base their gear-buying decisions on this meaningless data, and then wonder why the gear does not live up to their expectations.

A graph is much better, but even graphs can’t tell the whole story.

We live in an amazingly complex world!

Figure 2: The frequency response plot answers the question “What is the relative on-axis level change of the device-under-test regarding frequency?” (click to enlarge)

What’s The Frequency?

In Figure 2, the independent variable is frequency. The dependent variable is level. The frequency response plot answers the question “What is the relative on-axis level change of the device-under-test regarding frequency?”

For a device that produces the same level at every frequency, the plot would be a straight, horizontal line.

A real-world loudspeaker response is also shown. Some would consider a flat line response to be the best possible loudspeaker; however, a spectrum plot alone does not tell the whole story.

Now, let’s return to Figure 3.

Figure 3: At a glance, we can see that the loudspeaker’s on-axis directivity is increasing as a function of frequency (click to enlarge)

Again, the independent variable is frequency, while the dependent variable is the on-axis directivity.

The directivity plot answers the question “What is the ratio between the sound intensity on-axis to the total radiated sound intensity as a function of frequency?”

Q = 1 means that the device is omni-directional, where Q = 10 means that the intensity on-axis is 10 times the average radiated intensity.

Q = 100 means that the axial intensity is 100 times the average intensity.

Another way of describing the same thing is to use the directivity index, which is the Q rating converted into decibels with the formula DI = 10logQ.

It yields the same information in decibels, giving the loudness advantage produced by controlling the sound radiation.

DI and Q are often found on the same plot.

Turning our attention to Figure 4, once again the independent variable is frequency.

Figure 4: The impedance plot shows the opposition produced by the loudspeaker to current flowing from the amplifier as a function of frequency (click to enlarge)

The dependent variable is impedance.

The impedance plot shows the opposition produced by the loudspeaker to current flowing from the amplifier as a function of frequency.

A large peak on the curve means that less current is drawn at the frequency of the peak. This can happen at frequencies where the loudspeaker system is resonant, i.e. vibrates naturally.

Other frequencies require much more current to produce the same sound pressure level. Low spots on the curve represent frequencies where maximum current is drawn from the amplifier, i.e. where the amplifier is under a greater load.

The low values should be used when determining the required gauge of loudspeaker wire that should be used, or how many loudspeakers can be run in parallel.

Impedance is also required to calculate how much amplifier power is delivered to the loudspeaker, which in turn allows the loudspeaker’s power handling limits to be assessed. This is a good example of where a single number impedance rating (often called the nominal impedance) serves as little more than a guideline.

The impedance plot paints a much better picture of impedance and the other ratings that come from it.

Always remember to use specification sheets for what they’re intended – determining the suitability of a product for an application.

They are not a substitution for listening and measurement when evaluating products and should not be the final word in the buying decision.

A famous physicist once said, “The data on a spec sheet may be the best data they ever took or the only data they ever took!”

Pat Brown teaches the Syn-Aud-Con seminars and workshops. Synergetic Audio Concepts (Syn-Aud-Con) has been a leader in audio education since 1973. With nearly 15,000 “graduates” worldwide, Syn-Aud-Con is dedicated to teaching the basics of audio and acoustics. For more information visit their website.

Church Sound: Acoustics… It’s All About Signal-To-Noise

This past weekend I was at a Regency Ball with my family for an afternoon of exquisite ballroom dancing (think Jane Austen, 1812). The ballroom is located in South Bend, IN, in a restored old building called the Palais Royale. With its tall ornate ceiling and beautiful wooden floor, it’s one stunning place! 

At the ball, there was live music (piano, harp, and flute) and a “caller” who walked us through the dances. He used a handheld wireless microphone feeding a system with a number of ceiling loudspeakers placed throughout the space. 

The loudspeakers seemed to be placed well enough to deliver adequate coverage, and their overall sound quality was fine. However, the majority of the time I found it difficult to understand what the caller was saying. Why did this happen?

1) The room was reverberant; the hard smooth surfaces of the floor, walls and ceiling acted as great reflectors of sound.

2) There was a lot of additional noise created by the participants.

3) I hate to admit it but my hearing isn’t what it used to be. (Ah, you can’t beat getting old!)

While I was struggling to understand what the caller was saying, my brain was pondering the real problem. I recalled something my good friend Vance Breshears (principal consultant at Acoustic Dimensions) said a little over a year ago: “acoustics can be simply defined as signal-to-noise ratio.”
 
In the ballroom, I could hear the caller (the signal), but at times the people and reflections (the noise) were interfering with the signal. What could I do to reverse this ratio?

First, and I might add, simply, I began joining sets of dancers that were close to the caller. I figured if I was able to get close enough to hear the acoustic sound of his voice, the signal would be stronger than noise. 

Second, I worked to get myself positioned with “more mature” dancers, figuring they were probably having the same problem and would thus listen more intently and be quieter while the caller was talking.

My plan greatly helped, although being able to hear better did not mean I danced any better… :>)

I’m telling this story to get to this point: in your space, what is the noise? Is it the door at the back of the sanctuary that creaks, becoming horribly noticeable during the middle of the service? Is it the HVAC system that sounds like a wind tunnel when it fires up? Is it something simple, like the ushers forgetting to shut the doors to the lobby/narthex, allowing chatter and other unwanted sounds (footsteps, traffic noise outside the building, and so on) to drift in?

This Sunday, take a listen to the noise that interrupts the message (signal) and see what you can do to eliminate or at least minimize it. Sometimes things aren’t just about your system and mix.

Gary Zandstra is a professional AV systems integrator with Parkway Electric and has been involved with sound at his church for more than 25 years.

Thursday, March 29, 2012

In Need Of The “Big Hammer”

Many regional sound companies today have made the strategic decision to invest in smaller mid-size line arrays as their main speaker systems of choice (the “A” rig), as oppose to making the huge investment of owning a large-scale “big box” line array system. 

One of the reasons for this is a smaller box system is much easier to “scale up” for larger events, than it is for a large box system to “scale down” for smaller events. 

And we all know how painful it is to see that large scale system sitting in the warehouse waiting for that next big festival to put it out on (a few times per year), while most of the daily “bread & butter” jobs are utilizing the “B” and “C” rigs from our inventory.

However, not owning a “Big Hammer” rig in the tool box does present its own problems when trying to be the “go to” company in a given region. Cross-rentals are not always an option, and in some cases not desirable, even when readily available due to potential “competitor poaching” of clients.

So how can a company compete when the investments made don’t include the “big hammer” rig? This story illustrates one possible solution.

Starway Productions from Southern California is a moderate sized “full production” house (Audio; Lighting; Video; Staging, etc) that currently does not own a big-box system in their inventory. 

Their client base includes several large casino resorts, municipal and corporate clients and several concert and festival promoters (just to name a few). They decided to make their main loudspeaker system of choice the QSC WideLine 10 system (for most of the reasons stated above). 

And they have found over the past several years that they can easily satisfy over 80 percent of their clients without having to own a big-box rig. For those occasional artists whose rider demands an SPL capability exceeding 115 dBA at mix position, Starway would typically cross-rent a large box system from one of the big touring companies in the region. This got them through the event, but often cut deeply in to their profit margin. 

This past summer (2011), Starway completely designed and produced an outdoor “Summer Concert Series” for Pechanga Resort & Casino in Temecula, CA. – including the temporary venue, large stage, and all the technical production.

Many world-class artists were booked for the season, most of which representing a wide range of music genre, with a few of them requiring a “Big Hammer” system. Austin Hill, (Starway’s senior audio engineer) contacted me to see if we could come up with a system design utilizing their QSC inventory that could meet the added (SPL) demand of those few shows.

We decided to create a large “5-way” system for the main left-right hangs, which included 20 flown WideLine 10 (WL10) enclosures, 8 flown WL218-sw dual 18” enclosures as “LF” devices, and 8 x stacked WL218-sw dual-18-inch enclosures as the aux-fed sub-woofers on each side of the stage.

In addition to the main L-R system we also hung a 10 box WideLine 10 center array as a “vocal only” system to enhance the vocal clarity and point-source imaging throughout the venue. For the crowd of people leaning on the barricade down front, we placed 6 self-powered QSC K8 speakers spaced evenly across the lip of the stage.

In order to make each of the flown Left and Right systems behave as one coherent array, we changed a number of parameters in each system’s processing & deployment methodology.

First we flew the LF array so its vertical acoustic center matched the vertical acoustic center of the WL10 array, and carefully articulated the splay angle of the LF array to match the curvature of WL10 array.

Because the venue was only 175 feet front to back, I was able to keep the array shape to a very modest bend and maximize the potential sound power of the array using a combination “arcuate” into a slight “spiral” array shape. 

Using some custom link bars, we set the array curvature angles at .5 degrees for the top four cabinets; 1 degree for the next four cabinets; 2 degrees for the next seven cabinets; and 3, 4, 5 and 6 degrees on the last four cabinets.

Next, crossover points in the WideLine 10 array were changed to allow for the added LF array. WL10 normally utilizes a frequency shading topology between its dual 10-inch woofers, wherein we allow both woofers to go all the way down to its bottom knee (usually 80 Hz when used with subwoofers). 

At 130 Hz, the LF assist woofer begins to roll out at a gradual 6 dB per octave slope so in the upper mid range frequencies, only one of the 10-inch woofers is actually operating – eliminating off-axis time smear and nulling (cancellations) in the MF region. This gives WL10 the bass response of a much larger box (down to 52 Hz), while keeping the footprint of the box relatively small. 

However, the caveat to this topology is the MF device becomes the hardest working device in the enclosure, because it is doing double duty – MF and LF. So the idea was to take some of the strain off the MF woofer and let the LF array take on more of that low and low-mid duty.

Normal WideLine 10 high-pass and low-pass filters (absent contours), (click to enlarge)

With the MF 10-inch drivers positioned to the inside (on stage) of the arrays, and the LF arrays hung to the outside (off stage) of the WL10 arrays, we raised the high-pass of the MF 10-inch device to 160 Hz, and the LF 10-iinch device up to 100 Hz and kept all the low-pass filters the same. 

Then, we programmed the dual 18-inch LF arrays to operate from 40 Hz to 150 Hz acoustically using an 8th order Bessel low-pass filter at 110 Hz (See composite below).

Taking time arrival measurements dead center of the audience area (85 feet from stage) of each band-pass (left and right separately) we time aligned each within .02 mS of each other (including the center array, lip fills and ground-stacked subs). We allowed the center array to be 2mS ahead of the main left-right system just to put the vocals a little more “in your face.” This technique worked very well in the show as the vocals really stood out on top and were very clean.

Composite WideLine 10 high-pass & low-pass filters for the LF abd MF bandpasses, (click to enlarge)

The net result in this experimental pairing of these two separate arrays (into one 4-way system) was a significant increase in midrange clarity and headroom; increased bass extension and impact of the flown system before subwoofer use; and increased overall sensitivity and system output (total SPL). 

And having that much bass extension in the main arrays allowed us to use the ground-stacked dual 18-inch subwoofers as an aux-fed sub harmonic effect for those sources needing to go that deep. We limited the subs acoustical band-pass to only work from 28 Hz to 50 Hz, resulting in the low-end being very big & very tight!

(click to enlarge)

In summary, what we created was a big-box line array using smaller modular components. Some may say that we used an enormous amount of cabinets for a venue that size. However, this was done to meet an SPL requirement for only one or two acts for the entire season. 

When comparing a small box system to a large box system, it’s not the number of cabinets but rather the height of the array that determines its potential SPL output. The 20-box array we hung measured 16.8 feet tall, which is roughly the equivalent of a 10-box large format line array (i.e. – V-DOSC, VerTec 4889). That’s actually not that large of a system for a venue this size (4,500 seats) needing that amount of SPL and vertical coverage. 

(click to enlarge)

The cool take-away to this experiment for me was that this technique is scalable to smaller venues and smaller box counts - It’s making one “big hammer” out of a few smaller ones… Happy mixing out there.

Brian English is director of concert system solutions at QSC Audio (www.qscaudio.com).

L-Acoustics Introduces LA Network Manager 2 Software

Get more of the latest news from the 2012 PL+S show.

 
L-Acoustics is pleased to announce the release of the new LA Network Manager 2 software and its suite of interactive tutorials.

Completely rewritten from scratch on both network communication and GUI standpoints, LA Network Manager 2 features:

• An application-driven approach taking the user efficiently through the workflow steps of Setup, Tuning, and Live by implementing the tools required for each task into their own dedicated page
• A purely graphical interface optimized for tablet PC use and allowing for placing units and groups in the workspace in a way that reflects their location in the field
• Automatic discovery of the units connected to the network, including hot-plugged units
• New productivity tools: Unit Matcher, Custom Preset Builder and Preset Bank Builder
• Capability to assign units to multiple groups facilitating advanced system management and tuning strategies
• Two additional IIR PEQ in the Contour EQ
• Enhanced system monitoring with the Message Center and its log files

Wednesday, March 21, 2012

Recording: The Importance Of Space In A Mix

This article is provided by the Pro Audio Files.

 
We spend a great deal of time considering individual sounds in a space.

We prescribe attributes to the instruments and the players in order to organize our thoughts about the sounds and how they blend. We may often say a singer is “mid-rangy,” a snare is “ringy,” or perhaps the acoustic guitar is “warm.”

We do the same for microphones, pre-amps, compressors, and what have you. It is surprising how little time is spent considering the sound of rooms, reverbs, delays, and whatever other spaces are coexisting within our mix.

Considering that sound is defined by air vibrations within a space one would think the room would be held in equal importance to that which is resonating in it.

But, when entering a new space, how often do we consider it’s sonic characteristics? And more frequently, when building a mix, how often do we think of space(s) as its own sonic element?

Perhaps more often than we realize. After all, why do we spend so much time rolling through reverb presets trying to find the perfect one – when we seldom know what the right one will be? And why does a plate sound good one time, but a hall sounds better next?

Something instinctive is motivating these decisions. Like all sound sources, we are on some fundamental level listening for – and striving for – tone, rhythm, and coherence.

Reverb

The purpose of having customizable reverb is to find that which perfectly compliments the sound source – or the surrounding sound sources. We can pick and choose a reverb with a certain sound that highlights the tones or rhythms in our mix. And frequently, we will send multiple sound sources to the same reverb for the sake of coherence.

The complication comes in when there are multiple spaces present in the mix. After all, how can one element exist in two spaces at once? Or three? or, why is it that the choir sounds like it’s in a church while the lead vocalist sounds like she/he is in a concert hall?

Sonic Cues For The Listener

Of course, the end listener is not listening on such a discerning level. The end listener is only picking up on subtle sonic cues that either indicate the sound is coherent or disjointed. So our task is to lead the listener’s ear where we want it to go. Do we want a unified sense of space, or something surreal?

That’s our job as the artist, producer, or engineer. To orchestrate all the sounds and consider what feelings and emotions they evoke. They key word here being “orchestrate.” A random piling of sounds will certainly sound “unmixed” or perhaps more importantly, “ineffective.” Reverb and space are no exception.

Listening For Spacial Characteristics

The primary goal to understanding and sussing out any mix is listening. When listening to the drums, bass, vocals, strings, etc, perhaps we should also make a point of listening to the space in the capture. If you’re not used to listening to space then using a compressor as a listening tool with a fast attack and release and a low threshold will exaggerate the room sound in the capture.

Everything has spatial characteristics. A bass DI’d has no space sound – but that is still a spatial characteristic and must be considered. After all, if everything is close miked in isolation rooms, or DI’d, the capture is going to come out very dry, for better or worse (usually worse).

While listening to spacial sound, we are inherently listening to our front to back sound field. A DI’d bass is going to sound extremely forward while our drum kit miked from thirty feet away will naturally sound way back. This is a major advantage when organizing the image of our mix – as it can be recorded strategically to do the front to back work for us.

Tonal Cues

The trickier part of listening to space is the tonal cues. This is an immensely complex task, but can effectively be dumbed down into frequency response and “texture.”

This can be broken down into an even more fundamental question: Are the room sounds complimenting each other or clashing? A bright, open, Lexicon PCM 96 Hall reverb might sound fantastic on vocals, but if the acoustic guitar was recorded in a dark sounding, dense room, the two reverb sounds will clash (or at least sound incoherent).

While every mix is different, by and large this example will yield something that sounds “unmixed.”

Mix The Ambience

A brilliant colleague of mine named Gregory Scott turned me on to a unique but supremely effective concept. He said that one of the fastest ways to improve one’s mix is to “mix the ambience.”

I’ve taken this to mean mixing not just with the space sound(s) in mind, but actually take the time to get all your room mics, reverbs, and delays up front or in group-solo and mix them. Get the plate slap from the snare sounding like it belongs with the room capture on the guitar.

Or – if you have a surreal space – make sure it’s orchestrated in a way where the entire sense of space is working in the mix, or focus of the space moves in an evocative way (more on this in the next article).

Once all the ambience tracks are mixed start bringing in the elements that have the most space in them – drum OHs, and mid-distant strings for example – and focus specifically on their space and how it sounds with the other spaces.

Matthew Weiss records, mixes, and masters music in the Philadelphia, New York, and Boston areas. Find out more about him here.

Be sure to visit the Pro Audio Files for more great recording content. To comment or ask questions about this article go here.

Saturday, March 17, 2012

Loudspeaker Arrays: Ideas, Data & Solutions In Solving Horizontal Coverage Problems

A loudspeaker array is a collection of loudspeakers that is assembled to achieve a coverage pattern that cannot be achieved with a single device.

Arrays are most commonly implemented to achieve a wide horizontal coverage pattern from a position on or above the stage.

The “perfect” array would be a collection of loudspeakers whose radiation pattern was indistinguishable from a single (hypothetical) device that provided the needed pattern for the audience area.

Many attempts have been made to solve the horizontal coverage problem. These include:

• The “tight-pack” array ­ a collection of loudspeakers packed tightly together to emulate a single loudspeaker (Figure 1).

• The “exploded” array ­ technically not an array, but a group of devices that are separated by a sufficient physical distance large enough to reduce the acoustic coupling between the devices (Figure 2). Devices can be tilted at a downward angle.

• The “spherical” array ­ a group of devices with a common mouth distance to a virtual point of origin, placing them on the surface of a virtual sphere (Figure 3).

Figure 1, 2 and 3. (click to enlarge)

All of these side-by-side array topologies have merits if implemented properly. Let’s take a look at some facts and myths regarding the tight-pack and spherical arrays, and (hopefully!) provoke some thought about the horizontal coverage problem.

The balloon plots in this article were generated using EASE. They represent the approximate response of an array generated using the manufacturer-supplied EASE loudspeaker data.

Figure 4: Idealized radiation pattern. (click to enlarge)

Since real-world loudspeakers are inherently more complex than the EASE data representation, the simulations are “best case.”

The best-case response of any horizontal array could be described with the balloon plot of Figure 4. The plot is of three 60-degree horizontal devices arrayed side-by-side to provide a 180-degree horizontal radiation pattern.

NEED AN ARRAY?

Because a horizontal array attempts to achieve a wider coverage pattern than can be achieved with a single device, it makes sense to consider what such a coverage pattern would be useful for.

If the array is radiating equal sound energy to all points within its horizontal pattern, then even coverage is achieved only if all listeners in the horizontal plane are at a similar distance from the array.

Figures 5-7 show the audience planes that can be covered evenly with a side-by-side array.

We will proceed with the assumption that the goal of the array is to evenly cover one of these audience area shapes.

Note that if the array were tilted (i.e. above the stage), the audience plane would need to have the same tilt.

Such an audience plane is unlikely, so the “exploded” array is normally used this application.

Figure 5, 6 and 7: Optimum audience planes for a side-by-side array. (click to enlarge)

Figure 8 shows the physical conflicts that occur when a tight-pack configuration is attempted.

Figure 8: Ideal versus physically realizable devices. (click to enlarge)

If the acoustic centers could be reconciled physically, then a coherent wavefront could be achieved without the requirement of the sum of the individual radiation patterns being 180 degrees (Figure 9). Unfortunately, such a localized acoustic center is not possible for much of the spectrum in practice due to the required physical size of transducers that can radiate significant acoustic power.

It is necessary to de-centralize the components to a degree that doesn’t require the devices to occupy the same position in space. This process also moves the acoustic centers, and our “ideal” array is no longer ideal (Figure 10).

Figure 9: In a dream world…  Figure 10: The real world: our ideal array is no longer ideal. (click to enlarge)

The performance of a tight-packed array will depend on the degree to which the designer is able to reconcile the acoustic centers to a common point. Because a physical solution becomes more difficult with increasing frequency (shorter wave-lengths), the performance of tight-pack arrays will transition to that of a spherical array at some frequency.

Table 1: Maximum physical distance between acoustic centers of adjacent devices. (click to enlarge)

Table 1 shows the maximum physical distance bet-ween acoustic centers of adjacent devices that allow in-phase energy summation (less than one-quarter wavelength).

The spherical array moves the acoustic centers out from a common origin and uses a radiation pattern that minimizes the overlap bet-ween adjacent devices.

Figure 11 shows the ideal case, which would yield a “dead” zone in the overlap area. In practice, the opposite happens, since all loudspeakers spill some acoustic energy outside of their rated coverage patterns.

Figure 11: Spherical arrays move the acoustic centers out from a common origin. (click to enlarge)

The result is a “lobing” three-dimensional radiation pattern and an acoustic response riddled with comb filters at any single listener position.

It is interesting to note that the number of lobes in the radiation pattern is determined by the separation of the acoustic centers, not by the coverage angles of the devices that form the array.

Tighter patterns can reduce the level differences between the peaks and nulls, but they don’t reduce the number of peaks and nulls. Array performance is not judged by the absence of lobes, but by the relative level difference between the peaks and the nulls.

DIRECTIVITY DEVICES

Figures 12 - 16 (below) show the 3-D directivity balloons for several “real world” array configurations for frequencies in the voice range.

The geometric origin is 1 meter for each array, a distance that is great enough to remove the physical conflicts between the devices.

Figure 12 shows an array of small sound columns that have the typical broad horizontal pattern and controlled vertical pattern. The lack of pattern control produces significant lobing at all but the highest frequency considered.

At this frequency, the lobing becomes so dense that the response actually becomes smoother. Dense interference is a common technique used by sound system designers. As the lobe density is reduced (lower frequencies) the coverage becomes more uneven.

Figure 13 shows the resultant radiation patterns when the column loudspeakers are replaced with medium-format horns having a 60-degree nominal horizontal coverage pattern in the 2 kHz octave band. The coverage is much more even than in the previous example.

As with the previous array, these devices are positioned on the surface of a sphere by using a common distance back to a “virtual” physical origin. This arraying technique produces physically appealing arrays, but unfortunately does not compensate for the fact that the acoustic centers are not reconciled.

As such, significant lobing is present in the radiation pattern at the lower octave centers where the radiated pattern is wider than the nominal coverage.

Figure 14 shows the same configuration, but with the center loudspeaker advanced physically by one foot. This makes the array non-spherical, which (ironically) produces an improvement in the evenness of coverage in the 500 Hz and 2 kHz balloons.

Figure 15 shows the same configuration, but with the center device delayed electronically in an attempt to “compensate” for the 1-foot advance. This demonstrates that the acoustic center of a device is a physical characteristic and cannot be moved electronically. While a delay can certainly alter the radiation pattern of the array, it is not a direct substitution for the repositioning of a device.

Figure 12: Low-Q arrayed on a sphere. Figure 13: Arrayed on a sphere. Figure 14: Center loudspeaker advanced by one foot. Figure 15: Center loudspeaker advanced one foot and delayed .88 milliseconds. Figure 16: Large-format horn array with coaxial high-frequency section. (click to enlarge)

IMPROVING PERFORMANCE

Array performance can be improved by using devices whose directivity holds up to a lower frequency. This means a physically larger device.

Figure 16 shows the result of substituting large-format 60-degree horns for the medium format devices in the previous figures. The increased pattern control in the 1 kHz and 2 kHz balloons is apparent.

The bandwidths of these devices do not extend to 2 kHz, so the high frequency response was achieved with additional devices, coaxially mounted within the large-format horns.

Since using a larger format produces improved behavior, it is reasonable to expect that this improvement could be extended to lower frequencies if devices of sufficient physical size were used. Since the acoustic wavelength doubles when frequency is halved, the required size at 500 Hz would be twice that required at 1 kHz (8-foot mouth size!).

The wide horizontal coverage problem is one of the greatest challenges for the system designer. There currently exists no ideal solution, but there are certainly methods that work better than others.

Some conclusions of this and other studies are:

  • Pattern control is essential if the goal of the array is to emulate a single device.

  • Arrays of low-directivity devices should be avoided.

  • Arrayability is frequency-dependent. What works at one frequency may not work at another.

  • Spherical arrays are esthetically pleasing, but do not produce a common acoustic center.

  • Misaligning devices (either physically or electronically) may yield a frequency-dependent improvement in response.

  • Moving a loudspeaker produces a different result than delaying it.

  • Different array techniques should be used at low frequencies than at high frequencies (i.e. vertical line arrays).

Because architects and their clients insist on building rooms that are too wide to be covered with a single loudspeaker, the wide horizontal coverage problem will be an ongoing one.

This article should alert the designer and buyer alike to the caveats of the horizontal array, and help them identify designs that provide an adequate level of performance for a given application.

Pat & Brenda Brown lead SynAudCon (Synergetic Audio Concepts), conducting audio seminars and workshops around the world, and the leader in audio education since 1973. With nearly 15,000 “graduates” worldwide, SynAudCon is dedicated to teaching the basics of audio and acoustics. For more information, go to www.synaudcon.com.