How much of an improvement in the electroacoustic performance of an older loudspeaker system can be achieved by employing state-of-the-art processing and then optimizing the system using sophisticated measurement tools? I recently had an opportunity to test this idea when called upon by Walnut Creek Presbyterian Church (WCPC) in Northern California. The church is planning a major overhaul of their entire sound, video and lighting systems, but evaluations and improvements in rigging points and cat walks (etc) for the new systems would take approximately 2-3 years (possibly longer) and the church needed a short-term solution to their audio woes.

Their existing sound system is 12 years old. Yet, like so many other modern houses of worship, WCPC has expanded its worship style over the past decade to include contemporary music and dramatic productions. For this reason alone the loudspeaker system, which was designed for spoken word, not contemporary music, was showing weaknesses in providing the wider frequency response and spatial coverage that is now required.

Faced with the prospect of having to continue to use the existing system for a few more years, Jim Payne, the church's house system operator, inquired about rejuvenating the existing loudspeaker system and improving its base performance. His primary gripes included; high feedback susceptibility from the microphones used on the platform, problems with speech intelligibility, artificial-sounding music reinforcement and uneven coverage across the sanctuary.

GRAPH#1: Frequency response of FOH loudspeaker system before optimization.

Upon listening to the system, and taking a few measurements, I observed a significant bump in the low-end frequency response of the system that resulted in mushy, poorly defined bass. Likewise, there was a bump in the high-end response that made everything sound too bright, and produced a considerable level of audible hiss (see Graph #1). This initial evaluation also revealed that the layout and aiming of the high-frequency horns within the cluster was fairly good and that the cluster could provide adequate frequency response once it had been equipped with the proper processing tools, followed by a proper optimization. In choosing the new processing system we agreed to use a device or system that could be subsequently redeployed when the new system was installed.

A quick lesson in loudspeaker coverage and overlap:
Professional and credible manufacturers provide detailed information on product data sheets in form of polar plots measured at either one-octave or 1/3-octave (much better) points, as well as this same data in a form that may be used in software programs for system design, or room modeling. For example, a typical sound reinforcement loudspeaker rated for 60øH x 40øV coverage will typically provide wider coverage at lower frequencies and narrower coverage at higher frequencies than the 60-x-40 specification would indicate. However, the precise details of how a loudspeaker behaves at a given frequency are the data system designers use to create accurate predictions of how these products behave in a specific room.

Another key aspect to loudspeaker system design is that of controlling the interaction and overlap of loudspeakers that are assembled into a cluster (multiple speakers hung close together). Where this intersection occurs will depend on the loudspeaker height, the angle at which they are aimed downward and the angle at which they are aimed relative to one another. This intersection, or interaction of coverage is also frequency specific. Depending on the size of the horn(s), the loudspeaker's overall dispersion pattern(s), and the number and location of the speakers within the cluster, there may be intersection of certain frequencies, but not others. Although there are loudspeakers that do 'array' better than others, there are no loudspeakers made by man that can intersect with one another seamlessly or anything close to perfectly.

The key to designing a successful reinforcement loudspeaker system is to first select the best loudspeakers possible, and then determine where the intersection points may be and make adjustments to produce the least amount of destructive interaction. If this seems to be confusing or scary, it is ..until you've studied all that is involved and have done enough designs to obtain a healthy balance between the purely technical issues and experiential intuition. This is why loudspeaker system design is high on the list of the most difficult skills the consultant/designer is faced with, and why there may not be that many designers who can do this well.

The final chapter in this short walk-through on loudspeaker systems is on measurement and optimization. In those areas of loudspeaker coverage where it is not possible to prevent coverage overlap, it is possible to reduce the degree of destructive interference through alignment using digital processing. Careful application of delay, gain adjustment and, to a limited degree, equalization can reduce the consequences of this destructive interference. It has only been recently, within the past 10 years or so, that we have had the proper tools to measure such behavior, and to then treat it effectively. In a nutshell: each loudspeaker or group of loudspeakers that covers a distinct area of seating must be processed to align it with the adjacent loudspeaker or group of loudspeakers.

GRAPH #2: Frequency Response of 3 (of the 5) Existing JBL 2445 HF drivers, WCPC, August 2001.

We planned my two-day site visit to Walnut Creek Presbyterian Church accordingly: first, test the existing system components and (unless any are damaged) install and connect the new DSP processor. Next, conduct preliminary testing to see that everything has been wired correctly, and finally, conduct the electroacoustic measurement and optimization process. Due to time constraints, I chose to forego impedance measurements and hands-on driver inspection of the existing loudspeakers. Everything proved to be in good shape (see GRAPH #2) and then we removed the old processors and installed the new DSP system.

We chose a Yamaha DME32 Digital Mixing Engine because of its overall audio performance, variety of useful processing functions, very user-friendly GUI (graphical user interface) and its adaptability.

As shown in these photos, the existing loudspeaker cluster consists of five (5) high-frequency (HF) horn/driver combinations and three (3) single-15-inch woofers. The HF horns are JBL 'Bi-Radial' devices that provide consistent pattern control down to below 500 Hz. Although WCPC has effective acoustic treatments on its walls and floor, the sanctuary is a large enough space that there is the potential for problems from reflected energy off of the walls, should the loudspeaker system not be focused. The key (as always) is to restrict the sound system energy to only hit the seats and keep it off of the walls and ceiling. For this purpose the JBL HF horns were well chosen a dozen years ago, and remain aimed properly. However, the high- and mid-frequency coverage was uneven in level across the sanctuary due to the numerous user adjustments (primarily of gain) over the years. This would be easily corrected during the optimization process. Unfortunately, the three JBL woofers that had been installed at the top of the cluster were not aimed for even coverage. As a result, the woofers provide good low-frequency coverage to almost all of the seating areas, but there are distinct holes in the coverage at mid frequencies. Had we not been planning a complete system revamp in the years to come, it would have been logical to redesign this part of the loudspeaker system and possibly add more low-frequency (LF) devices.

Gain adjustments on the DME32 provided significant improvements in the even-ness of coverage throughout the sanctuary except in those areas that are undercovered by the woofers. We were also able to smooth out the frequency response of the system and significantly reduce the likelihood of feedback. Through equalization and delay adjustments, we also achieved a very noticeable improvement to the "naturalness" of the reinforced sound.

Following our two-day optimization process, WCPC held their Sunday services and afterwards Jim Payne's observations were very positive. He has found that he and the other sound mixers now have much more gain before feedback potential. And everything sounds clearer and everyone hears pretty much the same thing. Except, that is, for the choir and musicians on the platform.

Frequency response of choir monitor system, before (orange) and after (blue) optimization.

Frequency response of floor monitor loudspeakers, before (orange) and after (blue) optimization.

A second and equally long-standing problem at WCPC has been with their monitor loudspeaker systems. The choir never seems to be satisfied with what they were hearing (or cannot hear) and the musicians are continually asking for more level from their floor wedges. In looking at the position of the center loudspeaker cluster and knowing that the LF drivers were projecting considerable off-axis sound down onto the platform, I advised Jim Payne that this too would not be completely resolved until the new loudspeaker system could be installed. But the existing monitor loudspeakers were also ineffectively processed and I knew that we could improve things to some degree with optimization using the free channels in the new Yamaha processor. Each of the three mixes (choir, vocalists, instrumentalists) was rerouted through the DME32 and assigned both equalization and delay. Measuring from within the coverage area of each monitor system, the frequency response of each was first smoothed out and then aligned to the wrapped-around sound from the FOH loudspeaker cluster (sound coming off the back of the main PA). Following my visit, and the following Sunday service, the choir and musicians were visibly very pleased with the improvements. They all could now hear the lead vocals, piano and drums much more clearly and were able to hear well with reduced monitor levels. Out in the house and including the mix position, the low-frequency bleed from the platform and monitor system had been reduced and is now aligned to the FOH cluster. (See Smaart measurement graphs, left).

Although still very much in need of the forthcoming system renovation, WCPC will now spend the next few years with fewer obstacles to good sound.

Tom Young is principal consultant with Electroacoustic Design Services based in western, CT. He has just entered his 3oth year professionally involved in live sound and seldom wishes he still played bass or went into lighting instead. He can be reached at dbspl@earthlink.net.