Accidental Line Array

Introduction.

At the start of the twenty-first century the "in" design for sound reinforcement systems is the line array. After a slow start, all the major manufacturers have got their own variation on the theme which of course confers some advantage over their rival's systems with no disadvantage The main selling point for all these systems is their ability to approximate a true theoretical line source that radiates cylindrically rather than spherically. This reduces the rate of level reduction for each doubling of distance, as one moves away from the array, from 6dB to 3dB, greatly improving the "throw" of a system in large venues. Not to be outdone, I designed my own line array loudspeaker back in 1995 albeit for very different reasons. At the time I was unaware of any major manufacturer producing similar designs and it wasn't until I became internet enabled a few years ago that I became aware of the success of the Vdosc system and the subsequent development of similar systems by the other manufacturers.

The idea for the design originated from when I was working as one of the house sound engineers at a local venue. The installed system was beginning to show signs of its age and the owner asked me to think about what might be a suitable replacement. I thus thought about it and also the possibility that I might be able to design and install one of my own systems. Not long afterwards, the venue closed because the owners of the building sold it and the new owners wanted to use it for their own purpose rather than lease it. So apart from a couple of prototypes my design got no further.

The design.

I had never been a fan of the contemporary full range loudspeaker cabinets and their use when arrayed. The overlapping sound waves from the individual drive units created as many destructive interference patterns as constructive. Although it is possible to reduce these problems, in doing so the manufacturers would run into ergonomic, visual and manufacturing problems. With the available budget that the club had, and thus the type of equipment that they would be able to purchase these problems would have probably been ignored by the manufacturer.

The club was quite small with a capacity of approximately 400. It had a low ceiling and was wider than long (given the stage/PA position). Members of the audience were also free to move anywhere they wished and as such there were no dead spots. Given that everyone was on the same level vertically but at different positions horizontally, I wondered how I could reduce the loudspeaker system to a point source looking from a horizontal perspective. Column loudspeakers had been used historically but the traditional design would not provide a sufficient frequency range even if it could be built to produce the required sound levels.

As I was also involved with the design and construction of both hi-fi loudspeakers and the occasional studio monitor system, I was aware of the use of the mid-top-mid lay out used horizontally in a number of studio monitor designs (Quested, Dynaudio) and vertically in hi-fi speakers as introduced by Dr J D'apollito. Using the former layout but scaled up and with cabinets placed vertically on top of one another might produce the super column that I was looking for. The next stage was to draw a few sketches and do some rough calculations to see if the technique was worth pursuing.

Sketch1As a starting point a non-trapezoidal shaped box 1m by 0.6m by 0.35m was assumed. The dimensions limited the maximum size of bass drive unit to 12". If two such units were fixed at either end of a 1m front baffle there was only room for an 8" drive unit to fit between them. Whilst there are a number of coaxial 8" drive units on the market it was felt that the cost and sensitivity of such units was restricted for this design. That left three choices. Make the cabinet larger, abandon the design concept or move the bass drive units off the front baffle; this could be done by using a band pass type enclosure.
 
bassresponseThe next stage of the design was to calculate the response for various drive units. The conflicting parameters to consider are: frequency response, enclosure size, and limitations on the maximum SPL (sound pressure level) due to electrical and mechanical restraints of the drive unit. Calculations were done for drive units manufactured by Beymer, Eminence, Celestion, Fane, ATC, JBL and RCF. The best compromise was the Fane Studio12b, two of which would fit in the cabinet size specified, and achieve an output of 123dB over a range of 50Hz to 274Hz with no excursion power limitations and a ripple of less than 3dB (see diagram below).
 
Polar1The diagram shows the lobing effects of two point sources as a function of the wavelength between the sources. If any control over the directivity of the cabinet is to be maintained, which is a major feature of the enclosure, the upper crossover frequency must be at some frequency where the spacing between the drive units or ports in the case of a band pass enclosure is about 1/2λ , where λ = 1 wavelength of the sound. Although the responses shown in the diagram are obviously different the transition in response is gradual as the frequency increases. The actual cross over point can, therefore, be chosen to best compliment the response and directivity of the midrange section at some point between 200Hz and 250Hz (0.4λ and 0.6λ).
 
As a matter of interest calculations were done for a band pass enclosure where the rear enclosure is connected to the front enclosure with a port. As a "what if" exercise a simulation was done with the port for the front cabinet set at the area of the front portion of the cabinet and the tuning frequency set so that the port length would be close to zero. The response was similar to that of an ordinary bass reflex cabinet, which opened up the possibility of using a greater range of drive units, also given the omni-directional directivity of low frequency sound it should be possible to leave off the outer walls of the cabinet completely and have the speakers fire sideways. This could lead to problems of preventing damage to the drive units or grille during transit, but would be worth experimenting with. In the smaller of the two cabinets (assuming the rear volume remained as it is) the cabinet could be reduced in width by 240mm. Where cabinet size is less of an issue 15" drive units could be used which would give an extended bass output.
 

Sketch2After calculating thacoustic impedance of both exponential and conical horns and thThe band pass enclosure type means that there are now approximately 28 inches of baffle space left for the mid and high frequency units. This would allow either two 8" or 10" drive units to be fitted either side of a central horn. If the same principles are applied as used for the bass units, regarding directivity, the mid range units would be only operable up to about 500Hz and a small horn would most certainly not operate down to such a low frequency. The other alternative would be to horn load the mid range drivers using either a single 10" unit or two 8" units. In the enclosure considered the area of baffle available for the mouth of the horn would give an effective diameter of 0.34λ at the lower cross over frequency. This gives a reasonable loading to the drive units with little variation in the acoustic impedance.  When the cabinets are used in multiples the effective mouth area becomes even larger improving the loading of the drive units even further. eir theoretical response it was decided to use a conical design which allowed easier construction of a constant directivity type horn. Ease of construction coupled with better directional control was considered to outweigh the increased acoustic loading of an exponential horn. For the drive units I finally chose two 8" units that would face each other, a technique that I was familiar with.

The upper frequency limit of the horn is principally limited by the combined mass of the diaphragm and voice coil. The volume of air between the diaphragm and throat, however, acts as a compliance and although if chosen correctly can extend the upper response slightly it more often than not reduces the upper frequency limit of the horn. The lower individual mass of 8" drive units should give a wider operating range than 10" units. By careful design the response of the mid range horn should extend from 200Hz to 3500Hz.

Image2Initially it was envisaged that the high frequency horn(s) would be mounted coaxially in the mouth of the mid-range horn. With the possibility of a cross over frequency of 3500Hz it would be possible to use a number of cheap small horn driver combinations if cost was a major consideration. Where performance was more of an issue more expensive units could be used. While sketching possible layouts for the cabinet it became clear that fitting a slot horn in the throat of the mid range horn would be possible. This technique offers a number of advantages. Firstly the path difference between the mid and high drive units is less giving better time alignment and no lobing problems. Secondly the directional characteristics of both the mid and high sections are identical at the crossover point ensuring a smooth power response and thirdly the high frequency horn is fabricated into the cabinet saving the cost of a commercial unit. The idea for this technique came from the idea of making a vertical slot coaxial unit. I had always liked the sound of Tannoy coaxial drive units and had designed several successful cabinets using them. I had also been intrigued by an old Vitavox slot horn design (shown right), so combining the two techniques seemed like a good idea. The high frequency horn was made simply by cutting some inch thick board (diameter of the compression driver throats) to the required shape and sand witching it between two sheets of thinner board. The whole thing then fitted between the two midrange drive units.

The design was now looking feasible. Slightly more detailed drawings were created and the response of the mid horns predicted. These were modelled for some eminence 8" drive units that I already had and would use in the prototype, and also for 8" RCF drive units that I had considered for the "production models". The hi-frequency horn inside the mid-range throat would work down to 2000Hz in free air so I foresaw no problems with this part of the design and just went ahead with it. The prototype was loaded with some old JBL2420 drive units that I owned.

Whilst this is by no means a simple cabinet to construct, for anyone experienced in building loudspeaker cabinets should have no problem building such a cabinet. As well as considering it a good design from an acoustical point of view, the cabinet works well as a package. The design is simply a mid/high horn with the low-frequency enclosures fitted in the space that is normally wasted air in the horn box.

line array

The prototype cabinet was powered by some of the Mosfet power amplifiers described here. A 600W mono amplifier drove the bass and either side of a 350W/ch amplifier drove the mid and top. The crossover was also a home made one that was lying about in my workshop. The crossover frequencies were slightly wrong (250Hz and 2.5KHz) but close enough for me to dismiss the idea of building another.

Note the sophisticated computer measuring system. This was a Toshiba portable computer with a 386 processor and 4 meg of memory. An analogue - digital converter connected to the parallel port and the software ran in DOS. The speaker was later re-measured using windows based software Spectralab.

speaker testInitial measurements showed that the bass response was pretty much as expected, the mid horn worked better at low frequencies than I had expected, but the response was a little irregular at higher frequencies. Despite this the midrange sound was clear and coherent. The results for the high frequency response just revealed that in their old age the JBLs had lost a little bit of their top end. Despite throwing in the drive units that I had on hand, and using a crossover that had been designed for another system the cabinet sounded well. I was impressed by the detail in the mid range and the fact that even close up to the cabinet you could here the entire frequency range. Others seemed to be more impressed by the depth and clarity of the bass. The response seemed to go lower than measured. Despite the favourable comments no one was interested in the design because of what "other people" might think if they used it. It was just a little bit too different from all the sort of vaguely Turbo TMS4 copies that were being used.

For a couple of years the cabinet stood in my workshop as a pretty good system for playing tapes through. In 1997 someone contacted me with regard to building some, but by that time Fane had stopped making the studio 12b drive units and the order seemed dubious anyway. I recently moved house at which time the drive units were removed and the cabinet was disposed of.

 midhorn graphImage1