Line arrays were also popular in the 1950s and 60s because of the ability to provide excellent vocal range intelligibility in reverberant spaces.
Figure 1, Figure 2 and Figure 3 are excellent representations of high performance “vocal range” line arrays. These line arrays, like all vertically oriented sources in the past were, what could best be termed, limited bandwidth line arrays.
|Figure 1||Figure 2||Figure 3|
Figure 3 shows an Electro-Voice line array from the 1970s. It represents a relatively elegant solution to achieving high vocal intelligibility. It should be noted that the source separation of this design is roughly six inches, relating to a wavelength of 2.26kHz. The line array behaved very well up to that 2 kHz range.
It should also be noted in the Figure 3 that a high frequency horn was employed above that frequency limit in order to achieve appropriate extended bandwidth and fidelity up to and beyond 10 kHz. This is a classic embodiment of a limited bandwidth line array and as we shall see in this presentation, only recently have solutions been brought to the state of the art to enable line array technology to truly be full bandwidth and extend beyond the 10-15 kHz region.
Realizing a Full Bandwith Line Array
Full bandwidth line arrays are typically three way systems. The practice of dividing the band into 3 separate passes is done to enable the cross-over points to always be substantially low enough that the radiation from each pass exhibits wavelengths that are always longer than the physical device, or driver spacing. This is relatively easy to achieve for the low frequency section of any line array and is also easy to achieve for the mid-band section.
In mid-band sections the mid range devices are 6 inches in diameter to 8 inches in diameter. The crossover points are selected so that the device spacing is always small compared to the wavelength radiated. The problem for a full bandwidth line array systems is the high frequency radiation.
As mentioned earlier, historical line arrays were excellent in terms of low frequency and mid-band control of the pattern, but always suffered from polar lobing errors associated with the device space “B” being greater than the wavelengths being radiated. A 16 kHz wavelength is on the order of 3/4 of an inch and as a consequence device spacing must be comparable to those wavelengths or shorter, if possible. This was always a problem in the past because engineering techniques could not realize spacing closer than the driver diameters themselves.
Even with modern neodymium iron boron based magnetics, the diameters were always at least 4 inches or greater (for large format diaphragm devices). That spacing limited good performance to below approximately 3 kHz, obviously not a full bandwidth device.
As a practical example, fmax, the maximum high frequency control based on the relationship between the spacing of the devices b and the wavelengths is as follows. For base line arrays where we are interested in control up to 250 hz, the spacing needs to be at least 4.5 feet. This is relatively easy to do with 15 inch and 12 inch drivers and as a result the realization of bass frequency line arrays is very straightforward.
For mid-band line arrays, if we are interested in frequencies between 250 and 1,250 hz, the spacing needs to be 11 inches or smaller. Again, this is relatively easy to do with 6 inch or 8-inch drivers, and this is frequently the diameter of mid range devices in both large format and compact line array systems.
Figure 4 shows an Electro Voice Hydra™. This device basically takes the radiation of a compression driver and acts to produce both equalamplitude and equal phase sources at the front of the wave-guide. The full drawing in Figure 4 is 3 Hydras vertically stacked, thereby generating 21 “point source” radiating surfaces coupled to a horizontal wave guide with an included angle varying between 90 and 120 (model dependent).
Figure 5 shows a Hydra without the driver or wave-guide coupled. Each hydra has 7 output “slots”. The driver is coupled to the input side of the hydra and the 7 outputs are then interfaced with a horizontal wave-guide to produce the required horizontal included angle. The space b for a hydra is .826 inches, which equates to a wavelength of 16,434 Hz.
Again, it is always best for wavelengths to be longer than that spacing, so in this implementation, the Hydra presents excellent high frequency control in the 15kHz to 16 kHz range. The Array Show plot Figure 6 shows 21-point sources in a vertical orientation with the exact spacing provided by a hydra.
Realized Line Arrays/Horizontal Geometry Figure 7 represents two possible methods of orienting a full bandwith line array. The two methods are axis symmetric and axis asymmetric. The most common realization is that of an axis symmetric. It is the left hand drawing on Figure 7.
The high frequency section is in the horizontal center of the enclosure and is flanked by two mid drivers of 6 to 8 inch diameter and two low frequency drivers of 12 inch to 15-inch diameter (depending on individual realization).
One of the advantages of an axis symmetric design is that horizontal response is the same either side of the center axis.
Figure 8 slows a close up of an axis symmetric design. Of course one of the consequences for axis symmetry is that devices now become horizontal “arrays”. For most of this paper we’ve focused our discussions on vertical orientation of arrays, but it should be remembered that the same directional response characteristics exist for devices whether they are oriented vertically or horizontally.
There is a common mistake in sound reinforcement practice for people who normally understand that stacking devices vertically will control the vertical pattern to then stack devices horizontally in the misguided attempt to increase the horizontal radiation pattern. This is something termed array arithmetic.
In normal arithmetic, 40 + 40 + 40 will always equal 120. This, unfortunately, is not always the case with acoustics. In the same example, three enclosures stacked horizontally are usually done so because the array designer or the person developing the array has a desire to cover an included angle of 120 degrees (an example). The three 40 degree devices stacked horizontally will add to 120 degrees under certain conditions.
They will also add to 20 degrees when the wavelengths are comparable to the spacing between the devices. It should be remembered by all designers that stacking, whether the arrays are horizontal or vertical, will always narrow the pattern in the axis that the devices are oriented. This brings us back to the mid range devices and low frequency devices in an axis symmetric design. These axis symmetric designs are small horizontal arrays.
Figure 9 shows two eight inch drivers separated by a one-inch exit vertical slot for high frequency radiation. The two mid devices are oriented into a 90 degree included angle, but this spacing results in a horizontal array that exhibits the polar performance illustrated in Figure 10. When a cross over frequency of 1250 Hz is used, the response is basically 6 dB down at 30 degrees off axis generating an included angle of 60 degrees, not the 90 degrees desired by the designer of the product. This is the result of the classic “horizontal array” and will always occur when the crossover point is comparable to the device spacing. This, of course, can be eliminated by taking the crossover frequency substantially lower. Unfortunately, compression driver performance, in terms of mechanically generated distortion products and device reliability are severely compromised in the 700 to 800 region that is required for this type of device spacing. This is a classic trade-off seen often in acoustics where one parameter is optimized at the expense of a second parameter.
In this case, to achieve proper horizontal radiation and the desired included angle, the distortion, fidelity and reliability of the compression drivers are compromised; in order to produce proper fidelity, polar response is compromised.
An alternate approach is the axis asymmetric design also shown in Figure 7. In this design, there are no horizontal arrays. The trade-off, of course, is that the device voicing is not the same on the left hand side of the system as the right hand side. This, however, can be seen as a minor trade-off because the horizontal pattern is substantially improved and as a result, stereo imaging is enhanced. It has often been argued that the asymmetrical voicing produced by the axis asymmetric design is a design compromise but it can be seen as less of a compromise than that of the axis symmetric where the pattern begins to narrow or the sonic performance of the drivers is compromised because of using too low of a crossover frequency.