|Elliott Sound Products||PA Systems|
In the context of this article, sound reinforcement systems are referred to as 'PA' - PA (public address or sound reinforcement/ high power reproduction) systems are the life-blood of bands, DJs and many other performers. Regrettably, there is very little useful information about the systems themselves, or how they should be configured. One of the most common these days are commonly referred to as stick-boxes/ PA on a stick/ etc. ... boxes that sit on top of stands. These come in active (aka self powered) and passive versions, and can use speakers ranging from 200mm (8") to 380mm (15") diameter, usually with a compression driver and horn for the top end.
While these are easy to carry around and can make a lot of noise, they are generally a serious compromise. The latest offerings use plastic enclosures, and while these are much lighter than a plywood or MDF box (but only because the walls are so thin), they are often plagued by panel resonances because there is little or no effective internal bracing. However, when fitted out with a high sensitivity loudspeaker and a decent amplifier, they are often all that's needed for smaller venues. Adding a sub is necessary if a really solid bottom end is needed, because few of the boxes can perform well down to 40Hz, and many are struggling to get to 60Hz. Those that do get to 40Hz often do so only by applying bass boost (a peaking filter is common, similar to a single band of a graphic equaliser). This eats up available amplifier headroom, so the maximum output is never available.
By taking away the low frequency component, the amplifier is capable of a lot more output, because the bass boost circuit is no longer active. This relieves the headroom constraints in the amp, so crossing over to a sub at around 100Hz or so is well worthwhile.
Larger systems are also a compromise, but for different reasons. They are rarely self-powered, so require external amplification and active crossovers. One of the biggest problems facing those who use these systems is speaker failure, but the reasons are not well understood, and 'solutions' are often dangerously wrong.
Much of this comes about because few operators understand the reality behind the specifications of loudspeakers, and the concept of efficiency vs. power handling is almost never discussed. This is aided and abetted by marketing (dis)information, which is far more likely to cause confusion than impart any real information. The perception is that the only thing that matters is Watts. More Watts, better, lots more Watts, better still. When speaker system makers state 'output power' and give a figure in Watts, this is complete nonsense - I'd accept it if the values were around 10W (acoustic), but they actually refer to the input power, which determines the acoustic power only by taking the driver's efficiency into account. Making things a lot worse is the fact that many people who design some of the rubbish that's available now know little about the design of loudspeakers, less about designing reliable amplifiers, and less still about how their 'creations' are used.
The art of speaker (or system) design is knowing that there must be compromise, and knowing which compromise makes an audible difference and in what direction. The next stage is to assemble all the compromises in the one enclosure - if you get the mix right, you have a speaker, otherwise a large paperweight. Science assists the art - people made good speakers before the science existed, and continue to make bad speakers today - despite the science.
Let's look at a simple example. Two systems are set up side-by-side. One has an overall sensitivity of 100dB/W/m, meaning that an input of 1W will give an SPL (sound pressure level) of 100dB at 1 metre distance. Power rating is 100W maximum continuous average for this first example.
The second system is rated at 90dB/W/m sensitivity, but has a power rating of 1,000W - also continuous average. The question is ... which will be louder?
In theory, both will be exactly the same - they will be capable of 120dB SPL at 1 metre with full rated power. In reality, the 100W box will be somewhere between 3 and 6dB louder than the 1kW system, because at such a low average power there will be little power compression in the loudspeaker(s). All loudspeakers have to dissipate nearly every Watt from the amplifier as heat, which means that the voicecoil gets very hot indeed at sustained high power. For those that can (allegedly) handle 1kW, this is the same amount of heat as you'll get from a 1,000W radiant heater. As voicecoil temperature goes up, so does the resistance of the voicecoil itself, which increases the impedance, which in turn reduces the power obtained from the amplifier.
If only 3dB is lost to power compression, the amp power needs to be doubled (2kW) to restore the balance, but this makes the voicecoil get even hotter. This vicious circle continues until either the speaker fails or the amps have no more to give ... often both. A 90dB/W/m loudspeaker has an overall efficiency of about 0.62%, so with 1kW going in, only 6.2W comes out as sound - the remaining 993.8W is converted to heat. The 100W speaker (100dB/W/m) has an efficiency of roughly 6.2%, so 6.2W emerges as sound, and only 93.8W is wasted as heat. It is far easier to remove 94W of heat from a loudspeaker than it is to remove 994W - this should be immediately obvious. In both cases, the acoustic power is 6.2W, but it will only take a short time before the 1kW system shows power compression and reduces its output. Power compression figures for high powered loudspeakers range from around 4dB (very good) to as much as 7dB or perhaps more - this is not at all good.
In all cases, it's very common that the amplifier will be specified to deliver about twice as much power as the speaker can handle. The reason for this seems to be largely historical, but it is false reasoning in many cases. It is normal to describe the extra power as 'headroom' - some extra power to cope with transients so the amp won't clip. For many, many years it has been 'common knowledge' that speakers are damaged when amplifiers clip (distort). The conclusion is that if amps are big enough, they won't clip, and loudspeakers won't be damaged. No-one seems to have noticed that guitar amps clip much or most of the time, but the speakers usually don't fail. Sensible design for a guitar amp dictates that the speaker should be able to handle double the amp's rated power!
The next piece of common wisdom is that the use of a compressor/limiter will prevent the amp from clipping, and therefore will stop loudspeakers from blowing up. When both of these 'solutions' are applied simultaneously, this must surely fix the problem once and for all. Vast numbers of speaker drivers are destroyed because of this line of thinking. It's not (and never was) clipping that destroys speakers, it's sustained high power. An amplifier that's in full clipping (squarewave output) delivers twice as much power as the same peak-to-peak voltage of a sinewave, and that power doesn't change much regardless of the dynamics of the signal. It sounds seriously awful and speakers blow up.
With lesser degrees of clipping, the average power is still much higher than would normally be the case. Use of a compressor/limiter is only helpful if the attack time is short and the decay time long (at least a couple of seconds). When set up the way they are most commonly used, the compressor/limiter will simply maintain a (very) low peak to average ratio, and thus increase the average power delivered to the loudspeaker.
Contrary to popular belief, the speaker driver doesn't actually care if the sound is distorted or clean. Damage is caused because of the high average power, and if a limiter is used with an amplifier that has 3dB of headroom, it's probable that with some (most likely already heavily compressed) programme material, the speakers could be expected to handle close to the full amp's rated power for extended periods. There won't be any (amplifier) distortion, but the sustained high voicecoil dissipation and/or excessive cone excursion means that the speaker driver will have a short life.
While not commonly talked about, there are many additional complications created by power compression. Since the voicecoil gets hot with sustained power, this increases its resistance. This is understood, and it is this increase that reduces the power delivered to the loudspeaker and the subsequent SPL that emerges from the noisy end. The bit that no-one wants to talk about is the simple fact that the same resistance increase also changes the loudspeaker's parameters!
In particular, the speaker Qts is modified, meaning that the carefully aligned enclosure no longer works properly. The enclosure tuning can be further modified if the air inside heats up, and this can happen easily with many boxes, because there's either no airflow at all (sealed box) or the vents are incapable of creating a change of air within the enclosure. To put this into perspective, we know that 3dB of power compression is considered a very good result for modern high powered loudspeakers. This means that for all intents and purposes, the impedance has doubled.
Consider any passive crossover network. If done properly, they are designed for the actual measured impedance of the connected drivers. Now we have a conundrum - should we design the crossover network for a voicecoil temperature of 25°C, 80°C, higher, lower? It doesn't matter - at any voicecoil temperature other than that for which the crossover was designed, it is wrong! Bugger! In this respect, you can't win - so I suggest that any PA system intended for high levels must use active crossovers for all drivers. To do otherwise is asking for trouble.
This is especially important for horn compression drivers. A crossover network that's designed to be -3dB at 2kHz with an 8 ohm voicecoil will reduce the crossover frequency to 1.6kHz if the impedance doubles, and the compression driver will also be subject to an additional peak of 1.8dB at ~2.2kHz because the filter is no longer correct. This increases diaphragm displacement and can lead to failures. Likewise, the crossover loading is changed when the midbass driver's impedance doubles, so the crossover alignment is now completely wrong. Where the compression driver is in the same box as the midbass drivers, it can't be expected to remain cool - even if there is no power at all delivered to its voicecoil !
The above should be enough to make anyone think, but it gets a lot worse. In the next section, cone excursion is discussed. Cone excursion is determined by the box tuning and the power delivered to the loudspeaker. As the voicecoil heats up, its resistance goes up as discussed. When modelling the enclosure/speaker combination, the tuning frequency (based on internal dimensions and vent characteristics) is determined to get the optimum performance from the speaker. Again, should the box be modelled at 25°C, 80°C, higher, lower? Yet again, it doesn't matter, because at any temperature other than that for which it was designed, it will be wrong! Double-bugger!
The loudspeaker/enclosure combination modelled in the following section has issues - as will almost any design you can come up with. When the voicecoil temperature and resistance go up, the cone excursion also increases, so even less power can be delivered to the speaker before cone excursion reaches the danger zone. For the JBL 2241H, the tuning changes from being fairly flat down to about 36Hz (-3dB) to having a 4dB peak at 60Hz. If the operator applies EQ to make up for the loss of extreme bass, the speaker will be destroyed unless it is done with extreme care, and with full knowledge of the driver characteristics. We all know how often that will happen - exactly zero times for the life of the system.
None of the above considers the velocity of sound in air or air mass, both of which change with temperature and both of which directly affect the tuning of a loudspeaker enclosure. If the air inside the box is at a temperature of anything other than the design value, the tuning will be wrong, and cone loading may be found to be quite different from that expected. This can easily lead to greatly increased cone excursion and possible loudspeaker damage. There are so many variables that it will be impossible to even try to compensate. While it's theoretically possible to have computer monitoring of all parameters and apply compensation exactly as required, this would be a massively expensive undertaking and is unlikely to happen in the near future.
I mentioned 'stick boxes' earlier, and these are a prime case in point. Many dump a significant amount of the power amplifier's wasted heat into the cabinet, which simply accelerates the cascade of problems described here. The only saving factor is that almost all of these systems overstate the amp power, so the loudspeaker will hopefully have a reasonable safety margin - provided that the speaker's power handling hasn't also been overstated. Certainly, many of the more popular designs are surprisingly reliable - even when used by DJs, some of whom are renowned for their ability to convert sound equipment into scrap. Many other popular designs will simply blow up if pushed hard for extended periods, but there is no way to know which is which. Models change regularly, and a known reliable box today could easily become your worst nightmare tomorrow.
In short, it is worth considering a few major points for any loudspeaker/enclosure design ...
Needless to say, this covers only a fraction of the important considerations for a PA system expected to do anything more than amplify speech to a comfortable sound level. There are a great many other things that demand attention from the designer. Only by optimising a system for the desired parameters will a satisfactory result be achieved, and this invariably means that compromises must be made.
It must be understood that there is no such thing as a 'no-compromise' system - regardless of any marketing claim, they don't exist. The art of system design is knowing the difference between the really important and that which no-one will notice.
A (big) trap is cone excursion. Few people pay a great deal of attention to this, perhaps assuming that the equipment manufacturer will have taken steps to ensure that the cone travel remains within safe limits. From what I've seen of the available powered loudspeaker systems, the manufacturer has actually done nothing at all to limit excursion, and in some cases has actually incorporated bass boost circuits that ensure that linear travel will be exceeded. This causes greatly increased distortion, and exceeding the linear travel also increases voicecoil dissipation and reduces the cooling effect of the gap. Once the voicecoil has left the magnetic gap, instantaneous efficiency falls to zero, so every single Watt applied in excess of that needed is converted to heat. Without the proximity of the magnetic pole pieces, the voicecoil's temperature rise can be almost instantaneous.
In the case of separate enclosures, there is a minefield of traps for the unwary. There is a lot of information that is simply not published - not anywhere, by anyone. A perfect example of this is one of the more popular 460mm (18") bass drivers - the JBL 2241. These are very impressive drivers in all respects, and for a bass transducer they have a very respectable efficiency. JBL provides frequency response data with the 2241 installed in a 280 litre enclosure, tuned to 30Hz, and this appears to be a good alignment. Since the reference enclosure is described rather well, giving the port dimensions and box capacity, it's not unreasonable to assume that this will work well for normal PA duty for bands or DJ work. XMAX is a healthy 7.6mm, and XDAMAGE is claimed to be 40mm peak-to-peak (20mm peak). Power compression at rated power is 4.3dB - this means that the actual efficiency has fallen from 98dB/W/m to 93.7dB/W/m. To attempt to obtain the maximum SPL expected would require that amp power be increased by around 6dB (because there will be additional power compression at the higher power) - does it sound sensible to hammer a 600W driver with 2,400W?
There is just a tiny little problem, and it's not mentioned in the data sheet. If the 2241 is driven to its rated average power (600W) with an amp having a mere 3dB of headroom (1,200W), the loudspeaker will die. Cone excursion at 45Hz will be a very unhealthy 14.4mm peak (28.8mm p-p) - well in excess of the rated XMAX. At that frequency, the maximum instantaneous power that keeps the driver within its rated XMAX is ... 280W. You can imagine the damage that can be caused by a 1,200W amplifier driven to the onset of clipping. If this causes you some concern, then what happens below the 30Hz tuning frequency should really make you think ...
At a frequency of 24Hz, the damage limit of 40mm p-p is reached, and while this is not especially common with live music, it's easily achieved with recorded music - especially electronic music common for dance music and the like. Even with live music, it is necessary to prevent very low frequency 'transients' from getting to the speakers. These transients can be caused by switching on/off the phantom feed for a microphone, during setup for a variety of reasons, or simply by a bass player damping the strings with the palm of his/her hand. This can create signals as low as a few Hz, and a direct injection from the bass to the mixing console ensures that there will be plenty of level at 10-25Hz.
Figure 1 - JBL 2241H Cone Excursion at 1,200W Input Power
Figure 1 shows the cone excursion (taken from WinISD Pro) of a 2241H in the suggested 280 litre vented enclosure, tuned to 30Hz and driven with a 1,200W amplifier at full power. Quite obviously, something must be done to prevent anything below 30Hz from getting to the speaker, and for this reason, all vented loudspeaker enclosures used for PA work should have active high pass filters prior to the amplifier to prevent excessive excursion. In addition, to prevent driver failure, it is essential to know the power limits for each driver in the system, and ensure that amplifiers are sized accordingly. For the 2241, this indicates a maximum power of around 500W for each speaker - while this is still capable of exceeding the rated XMAX, it remains safely below the damage limit at all times.
Use of a high pass filter (such as the ESP P99 subsonic filter) is not only highly recommended, it should be considered mandatory. For this driver/enclosure combination, a 30Hz filter is perfect, and will eliminate excess cone excursion at very low frequencies - from any source. This is especially important if the system is likely to be used for DJ duties - turntable rumble and very low frequency feedback through the turntable suspension can generate awesome amounts of low frequency energy. However, even if no DJ will ever use the system, the filter should still be considered absolutely essential.
There's a further benefit too. By removing those frequencies that can't be reproduced at any useful SPL, the system will be a lot cleaner, with significantly less distortion. When the voicecoil leaves the magnetic gap, it cannot respond to any electrical signal until it returns. Without the magnetic field, the speaker essentially clips - just like an amplifier that's overdriven. Since very low frequencies aren't pushing the coil out of the gap at regular intervals, overall sound quality and sensitivity are improved and it will often be possible to get the same overall SPL with lower distortion and less power.
While I used the JBL 2241 as an example, the same principles apply to all loudspeakers. I used that one because I already had the details, and had modelled it previously. I also know that repairs to these drivers are common, and though few speakers are actually 'burnt out' due to overheated voicecoils, voicecoil and suspension damage are fairly common. Many other (cheaper) drivers are discarded when they fail, because recone kits aren't available and/or repairs are expensive. Experience with both construction and modelling using other speaker drivers tells me that a great many (probably most) of the drivers available today will show an almost identical trend. The actual figures will be different, but the principle is unchanged. One PA hire operator that I know of destroyed something like 30 18" drivers - a very expensive exercise to put it mildly. The combination of excessive 'headroom', highly compressed programme material and no high pass filters pretty much guarantees this result.
|Note Carefully: For reasons that remain totally obscure, for some time JBL reversed the polarity of their drivers. One expects that a positive voltage on the red (or + terminal) will cause the cone to move outwards, but JBL reversed this so positive on the black terminal causes the cone to move outwards. Incorrectly phased drivers in the same physical enclosure will be damaged very easily, because there is no loading on the cone. Newer drivers are phased correctly.|
Even drivers in separate boxes can be damaged if they are wired out of phase, especially if place near to each other. In this case, the bass output will be much lower than expected, so the operator will increase the power, likely to the point where the amp and speakers will be pushed to (or beyond) their limits. That is exactly what happened to the hire operator mentioned above. The power amps used had a bridging push button that switched half the subs out of phase. No-one noticed, so the amps were driven to the max, and one by one the subs failed until they were all defunct.
Loudspeaker manufacturers have all followed similar philosophies over the years. JBL (and before that, Altec) has always been a leader, with many of the smaller companies adopting the same general ideas. Users want to use more powerful amplifiers, so speaker makers produce loudspeakers that can handle more power, but usually at the expense of outright efficiency. Because no-one wants to have to transport really large enclosures, speaker makers modify the design to allow good (or at least acceptable) bass response in a smaller cabinet. Again, efficiency suffers because the cone must be made heavier to allow reasonable bass response in a small cabinet, so more power is needed.
About the only saving grace for subwoofers is that their impedance is much higher than nominal over much of their operating range. So, while the speaker may be connected to a 1kW amplifier, the actual average power is somewhat less than the maximum. If the impedance increases by a factor of 5, then power at that frequency is reduced by the same ratio. A 1kW (8 ohms) amp will deliver 200W into 40 ohms at speaker resonance. This is real and important, but don't assume that the reduced power at resonance reduces the cone excursion - the graph shown in Figure 1 refers to amp power, but in reality it's just based on the voltage from the amp.
While many makers seem to have broken the laws of physics if their advertising material is to be believed, this hasn't actually happened. What we have now are relatively small boxes that can produce a lot of SPL, but at a cost that many of the old-timers consider unacceptable (and yes, I'm one of them). No-one would ever claim that the PA systems used in the 1970s and even up 'till late last century were all 'high fidelity', but there were many systems that would cheerfully annihilate the vast majority of those around today. Some still exist, but because they are physically large and often time-consuming to set up, no-one is really interested any more. Systems used to have power ratings of around 1-2kW in total, but the use of horn loaded enclosures and efficient drivers meant that they were (extremely) loud, and if used properly very clean and punchy. Power demands were relatively modest, but it was always easy to get more than enough SPL in all but the largest venues with the systems of 'old'. Transport was difficult though, because of the size of the enclosures - which were also very heavy.
Today, power is cheap and transport is expensive, so small and light is preferred by most. A complete system that will fit into the back of a station wagon or ute is a better proposition for most operators than a system that fills the back of a reasonable size pantech truck. It is still a major compromise though, and the same level of performance cannot be expected from the smaller systems common now, regardless of claimed power. Remember that power compression becomes a major problem with any loudspeaker driven from an amp rated at more than ~200W (40V RMS into an 8 ohm load).
Now, I must warn readers at this point. Much of the remaining is not especially complimentary of systems, both old and new. It is important to point out that what I've written is mostly facts, but also includes opinions. Having worked in the industry for many years, it's impossible not to have opinions, some of which will inevitably be biased. I shall leave it to the reader to figure out which bits are biased. Just about everything written can be easily proved, but most information from a manufacturer's website can be considered to be only what it is ... information from a manufacturer's website. Despite many attempts to convince you otherwise, very little factual info will be found on these sites - there are exceptions, but it's not always easy to pick which is fact and which is fiction. One clue can be found - if the information seems to telling you things you'd rather not know (things that you are surprised that a manufacturer would admit to), then it's probably real. It must be noted that I'm not trying to sell anything to anyone (well, apart from my subsonic filter board), whereas that is the primary purpose of any website run by a manufacturer or distributor.
During research for this article, I came across some interesting material from one of the manufacturers of popular PA equipment. Mackie published an article WILL THE REAL MAXIMUM SPL PLEASE STAND UP?. According to the article, a common method of calculating the maximum SPL from systems is too bizarre to believe. If a loudspeaker's sensitivity is (say) 97dB/W/m and it is powered by a 200W amplifier (23dB above 1W), the maximum is 97dB + 23dB = 120dB SPL. For some utterly incomprehensible reason, it's common to add an additional 3dB to account for the 'crest factor' of a sinewave. What? This is complete bullshit! A sinewave has a crest factor of 3dB alright, but you can't just add that on to the SPL figure because it makes the specification look better (and that is all it does). The crest factor of a sinewave has absolutely no relevance to anything. I'm in complete disbelief that anyone would be so bloody-minded as to try to pretend that this is in any way real. At least Mackie takes the trouble to do a band limited pink noise test to determine SPL, rather than a nonsense calculation.
In reality, for any kind of meaningful calculation, you need to subtract about 1dB for each 100W of claimed power handling - if the manufacturer's data doesn't give a power compression figure. On that basis, the speaker and amp referred to would manage not 123dB as 'calculated' using the nonsense explained above, but around 118dB SPL (at 1 metre). This is much more realistic. However, some claims are quite obviously made up, and have no basis in reality whatsoever. Like PMPO, some figures are simply invented to impress, but are completely meaningless.
What you will not hear often but is nonetheless quite true, is that line arrays are all about sight-lines. Anything that gets in the way of the concert-goer's view of the stage cannot be tolerated, because it means some seats will have to be sold cheaply, or (horror of horrors!) not at all. Concerts 'in the round' are now popular too, and all pretense at stereo is gone. The whole system is mono, and everyone gets sound that is hopefully 'acceptable' (translation - "generally awful, but not so much so that people will complain".) I can no longer go to concerts, because the sound is so poor (and/or loud!) that I find the experience thoroughly hateful.
Almost all large concerts now are using line arrays for the PA system. These have become very popular, and even very small ones are available for smaller venues. While I know that many people will disagree, I consider the line array to be an unmitigated disaster in most cases. Those that I've heard all sound (often radically) different from each other, but they all share one thing - they generally sound bloody awful. Coupled with bizarre thinking about how they should be set up in the first place, the only ones I've heard so far that sounded even passable were in relatively small clusters (4 per side), and were situated high above the stage area. Contrast this with the glowing comments you may see elsewhere - a lot of people think that the line array is the best thing since sliced bread, and will wax lyrical about how they have solved all PA problems.
Bollocks! While there is certainly some real science involved (a much touted 'advantage' over earlier systems), for the most part the science has not made a system that's nice to listen to. Line arrays are fairly quick to set up, and may be much faster than the horn loaded systems they replaced. They are supplied with flying mounts, and even a large system can be hoisted up into position in a few hours. They are much smaller than the older systems, so are easier and cheaper to transport. They might even sound better than a horn system in some highly reverberant venues, but mostly they don't. They are comparatively inefficient, so it's not unusual for even a mid-sized system to be rated at 20kW or more. It is not at all uncommon for the 20kW of amps to be driven into clipping, making the relatively high distortion levels even higher.
Because everything is running at the limits, comprehensive monitoring is needed or loudspeakers will fail at an alarming rate. Many amplifiers now have remote monitoring facilities for power, temperature, speaker load, etc. for just this purpose. Some of the descriptions of, and explanations for, line arrays that I've read are just total rubbish - they are wrong in almost all significant respects. Manufacturers' literature is often no better - there is no science in having a marketing
puke executive write 'brilliant' sales copy. They want to sell the product, and don't give a rodent's rectum about the facts. While this may be less of a problem for very expensive professional equipment where prospective owners may want to test the claims before buying, gear that's affordable for bands to use is subject to little science and lots of hype. Something that's always worth remembering is ...
Wavelength = speed of sound / frequency (λ = c / f or λ = 345 / f)
Almost all line array systems require specialised equalisation, high slope crossovers (typically 24-48dB/octave) and some method of speaker monitoring to prevent overdriving the speaker drivers. This adds considerable complexity, and it's no longer possible for a few old-time 'roadies' (road managers) to set up the system. Some have dedicated software or spreadsheets to calculate the power distribution and array shape for a given venue. While this definitely real science, it's doesn't seem to have produced better sound for the most part.
The biggest single problem is that the effective line length varies with frequency. At 10kHz, even a couple of cabinets will easily exceed a line length of 10 wavelengths (at which point the line can conceivably be considered 'infinite'). At 1kHz, wavelength is 345mm, so the line must be at least 3.45 metres high to be considered close to an infinite line. At 100Hz, we need a line 34.5 metres long for the same effect, but needless to say this is usually out of the question. Tapered or shaded high frequency drivers can be used to restrict the effective or apparent length of the HF line as frequency increases. While this will reduce lobing and may prevent some of the problems, it is unlikely that the compression drivers used could keep up with the rest of the system.
For reasons most obscure, most designers claim that line arrays should face straight out, with no toe-in. This creates a hole in the middle of the venue where even a small head movement causes a most unpleasant phase effect, and also precludes any possibility of good stereo imaging. Because the treble 'line' is (very) long compared to wavelength, it delivers a lot more energy at middle distances than the midrange and bass, and tends to tear your ears off - a really hard, metallic sound that is utterly unrealistic in every respect. I would equate the sound 'quality' of most that I've heard with a cat farting into a milk bottle. In general, loudspeaker systems should always be pointing towards the middle of the listening space, preferably to a point about 1/3 of the room length. This depends on the room, and assessment has to be made on a case by case basis.
When speakers are angled so they point towards the front-middle of the auditorium, this is referred to as toe-in. By doing so, reflections from side walls are reduced, which in turn can help reduce room echo and reverberation. There's no fixed rule, but ideally, the centre lines of each speaker stack (or array) should intersect well before the middle of the auditorium, but even lesser amounts of toe-in invariably sound better than having the stacks/line arrays facing straight forwards. If you really want to mess up the sound and ensure that it's truly horrible, splay the speaker stacks as shown below.
Figure 2 - Correct and Incorrect Speaker Positioning
The simple trick of using toe-in has been used for hi-fi from the earliest days of stereo, and I can't think of any serious listener I know who would use a system set up with the boxes facing straight out, and none would tolerate the speakers being splayed. Regardless of anything that may be claimed, the vast majority speakers should always have toe-in. This isn't a new problem - people have been setting up PA systems without toe-in for decades, and for decades the systems have suffered from the awful 'hole in the middle' syndrome. It's not uncommon to see large line arrays set up with two columns facing straight out, and another pair splayed to cover the side areas. In a sense this is fair - everyone attending hears bad sound, so no-one is disadvantaged more than anyone else.
Unfortunately, there are a great many people around today who have never heard a decent sound system. Home systems that consist of little cubes and one-note 'subwoofers', MP3 players (with a cheap docking station perhaps) and generally pathetic live music systems have raised an entire generation of people who seem to think that what they have been listening to is 'good'. Anything that sounds different is likely to be considered 'bad' - and that probably includes dynamic range. Almost nothing on CD or live is free of massive amounts of compression - everything is compressed to within an inch of its life, and everything is the same volume. Real bass (below 40Hz), good stereo imaging and overall clarity are generally missing from all the music sources and playback systems that are available for reasonable prices, and even some big-name (and comparatively expensive) systems are woeful. It's pretty hard for anyone to realise that a PA system sounds like pox if that's all one has ever listened to.
If the reader is slowly getting the impression that I don't like line arrays, then I have managed to get the point across. In early times of PA (during the 1960s), the line array was all many of us had - typically 4 x 300mm drivers in a column enclosure. These boxes were almost always simply called 'columns' - the term 'line array' is much more recent. These columns didn't suffer from the same ills as the new versions - they had problems of their own though. It was not uncommon to find twin-cone drivers, with what was sometimes called a 'whizzer' cone - a small additional cone directly attached to the voicecoil former that made a reasonable effort to reproduce the high frequencies. While there were some issues with this approach, they were generally used only for the vocals, and managed to do a reasonable job at the time. Some makers (such as WEM in the UK) traditionally used a small horn as well as the cone drivers - this fixed some problems and created others. In some cases, a second set of column speakers was used, and was occasionally powered by a separate mixer/amplifier for drums and guitar. Most column speakers were rated at about 100-200W, and used 4 x 25W (or 4 x 50W) speakers, and amplifier powers of 100-200W were as much as could be economically obtained at the time.
One issue that line arrays have 'solved' is the ability to deliver acceptable sound at a consistent level to every seat in an auditorium. It's not about sound quality, it's about economics, sight-lines (sell more seats), setup and tear-down time, and making sure that everyone can hear the band. This minimises complaints (which are costly), and gives the best possible return to the promoters. That sound quality is no longer really considered is demonstrated simply. Look at the performance spaces that are being used now. The goal is to get as many seats as possible into the auditorium, including seats where one's view of the performance is insufferably bad. Stuck against a wall with a view of one end of the stage is not a way to see the performance, and it's quite obvious that the PA cannot possibly create a realistic stereo image if you only have one side of the speaker stack to listen to. I choose not to purchase tickets if that's the only option left - there's no point, and I'd rather spend the money on CDs.
Interestingly, the vast majority of comments about high quality sound come from the manufacturers, distributors and PA companies who have invested in line arrays, but very little from anyone else. Everyone I've spoken to thinks about as highly of the line arrays they've heard as I do - they are basically pretty awful. While they perform pretty much as claimed, this in no way should be taken to infer that quality is part of the equation. Line array makers like to point out how their products are superior to 'conventional' (i.e. horn loaded) systems because of their directionality, they fail to mention that the horn systems were often just as directional. They also may allude to the lobing problems of conventional PA systems, but completely fail to point out that the line array has its own lobing problems.
Figure 2a - Lobing Created By Unequal Distance From Listener To Drivers
You can't place two drivers handling frequencies up to 1kHz or so 500mm apart and not have lobing issues. At an angle that causes the distance between the drivers to be the same as one half wavelength, there is a deep null in the frequency response. Lobing is dependent on the distance between the drivers and the frequency, and causes a succession of peaks and nulls as the listener moves in front of the speaker system. Using toe-in can often help enormously, because when the listener is in a null zone for (say) the left stack/ column/ line, it is very unlikely that s/he will also be in a null for the same frequency from the right stack. Lobing of this nature is a fact of life, and we've always had issues whenever the sound source width or height approaches ½ wavelength at any frequency.
Some line arrays alleviate this problem to some degree by having the drivers as close to the HF horn as possible, sometimes angled inwards to form a simple horn as shown in the inset of Figure 2a. This is much better, and means that lobing is minimised. There is only a limited range of angle where both drivers are audible - however, the HF horn still needs to be crossed over at a relatively low frequency for this to be fully effective. Half a wavelength at 1kHz is only 170mm - it becomes readily apparent that there is lots of scope for problems. Because most systems use a fairly short diffraction horn, the ability to cross to the horn at a low enough frequency to prevent lobing is generally limited. Note that lobing is not limited to the horizontal plane - it also affects the vertical plane - lobing occurs whenever there is a path length difference between the listener and all audible drivers. Even if the high frequency line were absolutely continuous and can present a perfectly cylindrical wavefront, lobing will (and does) still happen with the horns as well. Some arrays manage this reasonably well, others don't.
There is almost an infinite variety of line array box layouts from almost as many manufacturers. Being the 'flavour of the month', any PA operator who doesn't use them is likely to miss out on work, because almost everyone has been convinced that this is the only way to go.
This is patently untrue and misleading, but it's extremely difficult for any operator to convince a client to use his/her ears and choose the system based on merit. It's even harder to convince a promoter, since the main thing that drives their agenda is the financial return. Most neither know nor care which PA sounds the best (or even better than something else), they are going to allocate the job based on a basic specification and price. A system that takes longer to setup (or horror of horrors - blocks the view of some punters so certain seats can't be sold) will never get a look-in, regardless of how good it might sound.
I've not had the opportunity to mix a live band through a line array, but I suspect that it would be possible to get a good sound from an average size array. If sound quality is the goal, then other factors must be sacrificed - this is the ever present rule of compromise. Sound quality can almost certainly be optimised, but at the expense of comparable SPL at every seat. Some seats simply cannot be used if sound quality is the target, because they are too far off to either side of the stage. It is essentially impossible to provide genuine high quality audio for any listener who is not between the speaker stacks.
Much was made about lobing in the above, but as noted, it happens with all PA systems that use more than one loudspeaker. The simple fact is that only a point source is free of lobing, and this cannot be achieved with any driver that exists. What is a point source? This is a theoretical sound source that is very small compared to any wavelength within the audio range. All sound is radiated from this one point in space. In the physical world, the point source radiator does not exist.
If the system consists of a single stack on each side of the stage, it is almost possible to avoid lobing, by crossing over each driver before its dimensions become greater than ¼ wavelength. This is easy enough for low frequencies, but becomes progressively harder as frequency increases.
High frequency horns suffer from lobing, because they are invariably larger than one wavelength at anything above a few kHz. Since a single stack of single drivers can never produce enough SPL for even a medium sized indoor gig, it is necessary to use more speakers. As soon as additional drivers are introduced, lobing becomes an issue - there is no answer, unless there is only one member of the audience, and that member is nailed to the floor. I cheerfully accept that this is not generally in anyone's interests (especially the poor bugger nailed to the floor), so we have to accept that ...
As already pointed out, to some extent lobing can be mitigated by using toe-in for the PA stacks, and this is necessary (and works) regardless of the type of system. We also have to accept that audience areas will inevitably spread beyond the stage width, so only a relatively small number of punters (patrons) will hear optimum sound quality. Likewise, it is these same punters who have the optimum view, and it is unrealistic to expect otherwise. Big video screens give vision to those who can't see the stage properly, but there isn't much that can cure the audio problems.
Many large concerts use delay stacks - separate PA systems further back into the audience areas that have the audio signal delayed to match the distance from the main PA. For example, if these stacks are 100m from the main PA, the signal is delayed by 290ms so the sound from the main PA and the delay stacks arrive at the same time - to a listener who is further back from the stage than the delay stacks (say 120 metres from the main PA). It's important that delay stacks have very little radiation from the rear of the boxes (which are towards the stage), and if subs are used these really do need to be directional. While folded horns are probably the best, active pattern control by means of additional drivers driven out of phase will be the method of choice these days.
As a mental exercise, it's worth thinking about the effect where a listener is slightly off axis to both the main and delay stacks, but can hear both clearly. What will be the effect if the time difference between the sound arriving from each is around 30ms (that's a path length difference of only 10 metres)? If you've not heard it, the effect is best described as 'interesting' - not quite an echo, extremely poor articulation, and very odd frequency response.
The simple reality is that it is unrealistic to expect that everyone in a venue will hear perfect sound, unless every punter is handed a pair of headphones as they come in. While attractive from a 'perfect sound' perspective, it not an idea that's likely to find favour for very large (or any) concerts . Once we look at the problem from a realistic perspective, the line array becomes a little more attractive, however, it is still critically important that the effective line length vs. frequency remains relatively constant. The biggest problem (referred to at the beginning of the Line Array section), is still at that critical distance from the array where the sound is decidedly 'top heavy' and tries to rip your ears off. This happens because the line array is very long compared to wavelength at high frequencies, but is progressively shorter as frequency is reduced.
To some extent, the standard 'J' curve that's applied to line arrays will ensure that this effect is minimised, so those who are relatively close to the system hear (at most) perhaps two sets of boxes in the array, and the number of audible boxes increases as one moves further away. Whether or not this actually solves the problem is up to the skill of the operators and those who install the system.
In essence, there is no ideal system - every approach has problems, and it's up to the sound engineers and promoters to determine the ideal system for each venue. Present thinking is that line arrays should be used for everything, but this is simplistic and unrealistic. Many venues would be served far better by a traditional horn system.
The column speaker was popular for quite some time. Larger ones like those described above were the most common for live performances, but for popular groups with lots of screaming fans, it was inevitable that no-one actually heard the band because the systems were unable to achieve the SPL needed to drown out the screaming. This was highlighted in 1964 during the Beatles tour - Sydney Stadium could accommodate 12,000 fans, but no-one had a PA system that was big enough. The entire PA system consisted of a couple of column speakers and mixer/amplifier that was probably no more than 120W or so. These systems were the mainstay of nearly all bands and even some major tours all over the world until the late 60s and early 70s, when things started to change. It's worth noting that column speakers are still used sometimes, especially where only speech reinforcement is needed. Churches commonly use column speakers for sound reinforcement, because they are reasonably directional.
It was well known by the 1950s that columns cause lobing and interference patterns in the vertical plane, so to minimise this some columns were 'tapered'. Not in the physical sense, but electrically. The centre speaker would be full range, and each driver above and below was driven via a low pass filter, continued as necessary depending on how many drivers were used. This made the column appear to have the same acoustic length (in wavelengths) for the frequency range of interest. Doing the same with line arrays might solve some of the inherent problems, but the high frequency horns and drivers are almost certainly far too wimpy for just one or two of them to be anywhere near loud enough to be heard.
Interestingly, there was very little cross-discipline activity in the early days of sound reinforcement for 'rock' bands. Extremely powerful audio amplifiers were in common use for AM radio broadcast, but none was ever used to drive loudspeakers (other than for testing, research or fun). No loudspeaker existed that could handle more than a few Watts, so maximising efficiency was very important. Until the late 1960s, none of the cabinet designs used for movie theatre sound reproduction were even considered for live sound. Each of the various audio fields tended to keep strictly to itself, so no-one learned much from anyone else. At that time, most amplifiers were valve (vacuum tube) based, and the maximum power that could be obtained from a portable system was about 200W. If more power was needed, slave power amps were sometimes used to drive more speaker cabinets. While more powerful amps existed, they were fairly uncommon.
Many concerts in the late 1960s used multiple columns powered by multiple amplifiers. This created a distributed source that did very interesting things to the sound, depending on where you were standing in relation to the speaker arrays. None of this even approached high fidelity, but the excitement of a live band or concert usually made up for the poor sound quality. With popular groups, the fans would still easily overpower the PA system anyway, so sound quality wasn't much of an issue. In this respect, little has changed.
5.1 - Horn Systems
The new systems that emerged in the 70s typically used fully horn-loaded designs. W-bins were adapted from movie theatre designs for the bottom end - the 'adaptation' was to make them small enough to move around, but that reduced their bass response. The W-bin was a folded horn bass speaker, and was much loved by a lot of bands because of the great chest compression it produced with the kick drum. Most struggled to reproduce anything below around 70Hz at significant sound levels, but this was better (louder) than anyone had heard before. There were various horn loaded boxes used for midrange, but like the folded horn W-bin, most were adapted from (mainly) Altec Voice of the Theater™ designs, as well as various JBL and RCA theatre systems. In many ways this was unfortunate, because the old theatre systems were engineered mainly for speech clarity in a theatre, and were based on the most efficient loudspeakers available. Power (and audio bandwidth) was severely limited, so the loudspeaker had to make up for the lack of electrical power. Fidelity was usually pretty woeful in theatres at that time (and in many cases still is), but if the audience could understand the dialogue then that was the best that could be hoped for.
Many quite large concerts (such as Woodstock in 1969) used remarkably little power. The Woodstock system is said to have used 10 x McIntosh MI-350 mono valve amps - a total of 3,500W. There is some disagreement on this, and surprisingly little real information. Many of the early (even quite large) systems only used about 1500W in total, and this was generally far more than could be obtained economically from any valve amps that were available at the time. Relatively large transistor amps became common in the late 60s (the Crown DC300/DC300A was rated at 150W into 8 ohms, but gave closer to 200W). These were only possible after various semiconductor manufacturers had perfected power transistors that were capable of handling significant voltage and current. It wasn't until around 1970 that these new devices became available at prices that mere mortals could afford, but once their price came down to something tolerable, things changed forever. Before this, the only (cheap) readily available high power transistor was the venerable 2N3055. In the early 70s, I built guitar amps and (mainly column) PA systems, and the only high-power transistor I could get at the time was the Solitron 97SE113 - now long gone, but not forgotten. The release of the Crown DC300 power amp in 1967 (closely followed by the DC300A, Phase Linear 200, 400 & 700 and a few others) signalled a new opportunity - plentiful power.
There were issues faced by these early systems that were not understood by many of those who put systems together. The theatre systems were engineered for particular drivers, but few people ever made changes to the design to suit the more powerful loudspeaker drivers (which behaved very differently) that became essential to get the SPL needed. As a result, many of the horn enclosures used simply didn't work properly, but they made up for any lack of fidelity by being far louder than anything that had come before. Boxes such as the Altec A6 or A7, or the JBL 4560 were stacked side-by-side with radial, sectoral or multicell horns on top. No-one seemed to notice that this arrangement caused huge phase and frequency response anomalies. Compression drivers were often used in pairs on a 'Y' throat adaptor to get more SPL (which usually didn't work at all well). Even today, many people don't seem to realise that a compression driver on a horn can only achieve an undistorted SPL that is based on the peak pressure at the throat. Small throats are necessary for good high frequency reproduction, but will have problems if driven too hard. A 25mm (1") throat horn simply cannot go loud enough before serious distortion if it is expected to keep up with perhaps a pair of high efficiency 380mm horn loaded midbass drivers. The reason is largely that air has a non-linear relationship between pressure and volume, so adiabatic compression and rarefaction can only approximate a linear function over a very limited range. A larger throat allows higher SPL, but reduced extreme high frequency response - one of many compromises.
|For what it's worth (and because you'll find very little on the Net), the maximum acoustic power into the throat depends on several factors. First is the relationship of actual frequency to the horn's cutoff frequency. As the ratio of f/fc (frequency divided by cutoff frequency) increases, so does distortion for a given acoustic power per unit of throat area. A sensible upper limit for throat acoustic power is around 6-10mW/mm², meaning that a 25mm (1") throat should not be subjected to more than 3-5W. A 50mm throat can take 4 times that power, or 12-20W acoustic (see graph [ 1 ]). The amount of acoustic power that can be accommodated decreases as frequency increases. For horns intended for operation from (say) 800Hz and above, the normal rolloff of amplitude with frequency (as found in most music) means that the acoustic power also falls with increasing frequency.
If the conversion efficiency of a compression driver is (say) 25%, this means there is absolutely no point supplying more than 20W (electrical) to a compression driver on a 25mm throat, or 80W for a 50mm throat, allowing for a sensible distortion of 2%. Past a certain limit (which varies with frequency vs. horn cutoff), supplying more power creates no increase in SPL, but simply creates more and more distortion. The maximum power must be reduced as frequency increases. CD horns require HF boost, so can easily be pushed much too hard at high frequencies, resulting in greatly increased distortion.
Quite obviously, any horn that has a small throat must have limited power capability, and providing amplifiers that are (much) larger than needed for 'headroom' is a completely pointless exercise. It is both convenient and accurate to consider the effect as 'air overload'.
According to a technical note from JBL [ 2 ], the situation is actually worse than the graph shows. A 200Hz horn at 10kHz can readily generate 48% second harmonic distortion, with as little as 2.5W (electrical) input - a mere 0.75 acoustical Watts. As noted in references 1 and 2, this information was first determined in 1954, but over time seems to have been lost. As you can see, I'm determined that this will not happen.
With the top end being handled by horns with compression drivers, the next question was "which horn?". The well heeled would often use multicellular horns while others used either sectoral or radial horns. In some cases you'd see a combination of different sized (and different types) of horns, which may or may not have been crossed over at different frequencies, and may or may not have been compatible. All worked well enough, but the multicell horns still have a place in the hearts of the many who ever used them - there is just something almost magical about the multicell that no other horn design can match. A web search reveals that there is great confusion in some quarters - some seem to think that a sectoral horn is multicell - no it isn't - they are quite different. The biradial horn (sometimes known as the 'bum horn' because it looked rather like someone's backside) saw little use for live sound. This was an attempt at what became known as CD or constant directivity horns. While a good idea in theory, they usually don't load the driver properly. This means that the compression driver must be de-rated or it will fail due to over-excursion. As noted above, CD horns require high frequency boost to maintain flat response, and this can lead to excessive distortion at the higher frequencies.
These high frequency horns actually came in many different forms. Sectoral, radial, multicell, diffraction, horns with acoustic lenses, 'bullet' tweeters, ring radiators - the list is long and diversity is great. Materials varied too. Aluminium castings were common, but many radial horns were timber, and later came moulded fibreglass and various other plastic materials. Every manufacturer claimed to make a 'better' horn than the competition. In some cases the difference was clearly audible at typical showroom demonstration levels, but at 120dB SPL, no-one could tell the difference.
In many cases, people used to refer to horns used with compression drivers as being either 'long' or 'short' throw. The theory was that long horns had greater directivity, so could reach the back of an auditorium whereas a short horn could not. This was mainly nonsense, because the directivity is determined largely by the shape of the mouth, and the length (and mouth area) determine the lowest frequency at which the horn can provide proper diaphragm loading. Still, it was a myth that was almost impossible to get rid of at the time, and it still persists. The horn arrangement that gave the best control of directivity was always the multicell, but they were always the most expensive of the many different types. Essentially, one can classify almost any horn as 'long throw', because they have controlled directivity. For good dispersion at close range, a horn acoustic lens (preferable - mostly made by JBL) or diffraction horn (not so preferable IMO) is a better proposition. Line arrays generally use diffraction horns.
Figure 3 - Altec Multicellular Horn
Some systems used JBL slot, bullet or ring radiators for the extreme top end, because the 2" throat compression drivers don't offer much above 8kHz or so, due to diaphragm break-up at higher frequencies and high distortion caused by excess power in the throat. Many people felt that the extreme top end was unnecessary, because at the kind of SPL one gets at a live performance, one's ears can no longer hear the last octave anyway.
While the horn loaded cone loudspeakers have almost vanished these days, 2" compression drivers and exponential horns are still fairly common. They are even used in some of the more powerful plastic 'stick box' enclosures, although most use 1" compression drivers. The horn is simply moulded into the enclosure, and while economical, they lack proper bracing and damping, and most are too short to use down to 500-800Hz as used to be common. Because almost no-one uses horn loaded midbass boxes such as the A7 or 4560, a significant loss of efficiency is experienced, which requires more amplifier power and all the ills that I referred to earlier. Many other horn loaded midrange boxes were used too - dual 12" horn boxes were common, and these typically worked down to around 200Hz or so. There's some useful background info on horns and drivers at the LenardAudio website. Indeed John Burnett (from Lenard) and I used to make horn loaded PA systems, using fibreglass enclosures (and horn flares) for the midrange boxes. The bottom end was handled by folded horns, which were ideally used in groups of four to get sufficient mouth area.
Larger systems in the 70s were generally fully horn loaded. The top end was handled as described above, and the bass and midrange were handled by a variety of different systems. In most cases, each had its loyal followers and often equally vocal detractors. Very few of these systems were actually designed for the loudspeaker that was to be installed in them, so performance often varied widely, even though from the outside they looked like any other of their ilk. Some of the boxes could be described as midbass - intended to handle both bass and midrange, but in many cases doing a poor job of both.
Because the enclosures were modifications of designs that were used for movie theatre systems, they often did not perform as expected when used with a high power (perhaps 150W or so) loudspeaker. In most cases, any deficiency was simply ignored. The boxes had been built, speakers installed, and the system went into service - warts and all.
Figure 4 - Altec A7 With Sectoral Horn
Figure 5 - JBL 4560 Midrange Box
The typical enclosures used for midbass varied. There was the Altec A7, another that was commonly known as the 'Roy' box, both horn loaded. Another popular enclosure of the day was the JBL 4560 - a single horn loaded 15" driver in a (kind of) vented enclosure. I must have seen plenty of Roy boxes, but unfortunately can't recall any details - a Web search indicates that they used 2 x 12" drivers and may have used a conical flare, but information is scarce. There were a lot of other designs as well, many of which were obscure even then, and most have passed into history now. Many of these were variations on the ones listed, and there were plenty of people making horn systems at the time.
Bass was most commonly handled by W-bins. These were made by several major manufacturers (Altec, RCA, etc.), but were quickly copied. The typical speaker complement was a pair of 15" or 18" drivers. Very few actually reproduced 40Hz, because the flare length and mouth size are simply prohibitive at that frequency. However, when used with two per side (or more) they usually managed to be able to deliver very high levels at around 70Hz or so - just right for the kick drum.
Figure 6 - General Layout of a W-Bin
Figure 7 - Cerwin-Vega Folded Horn
Nearly all folded horn boxes used straight sections, with the average expansion being (more or less) exponential. These boxes were big, very heavy, and difficult to move around - although they were still much smaller than those used in large movie theatres!. However, those who've never heard them in action would find it a jaw-dropping experience. A huge amount of power just isn't needed when you have an enclosure that boosts the efficiency to around 106dB/W/m with the right drivers installed. Coupling a portable transistor radio to one of these horns would have SWMBO and the neighbours yelling at you to turn it down in short order ... from as little as 250mW of input (and I know this from personal experience). Folded horns weren't all of the 'W' shape though - a great many bass horns used a single flare, as shown in Figure 7.
As with HF and midrange horns, there was a very diverse array of designs. Some worked very well, and some were only marginally better than a direct radiating loudspeaker. In nearly all cases though, the speaker had better protection from mechanical damage caused by over excursion than any direct radiating design. The majority of the systems I used, built, or helped design/build were exceptionally reliable. Although amplifier power was very modest by today's standards, these systems were all easily capable of exceeding the maximum SPL allowed in most venues (some of which used a 'traffic light' SPL cutout - if the red lamp was on for more than 10 seconds, the stage power was cut!). In the majority of band's PA systems of the 70s, it was almost unheard of that any loudspeaker would be driven much beyond 200W, yet these same systems were considerably louder than anything available now with the same power rating.
If you are contemplating using a bass horn (of any design), the use of a high pass filter should still be considered mandatory. While the rear compression chamber of folded horns restricts cone movement below the cutoff frequency, there is still wasted power and more excursion than may be desirable. Midbass horns (such as the Altec A6/7 or JBL 4560 designs) absolutely require the filter, as the cone is unloaded at low frequencies, so cone excursion can easily reach dangerous extremes.
It's also worth noting that a folded horn presents a relatively benign load to the driving amplifier. This is good, because it means that the amp's internal protection circuitry is unlikely to operate. Many amplifiers, both old and new, famous and infamous, have hyperactive protection circuits (examples are some Yamaha and Phase Linear amps, Bose, etc.). When these operate the audible result is usually very nasty indeed - much worse than clipping distortion - see VI Limiters in Amplifiers for more. An impedance that is largely well above the nominal rating means that the amp has an easy time, reducing wasted heat in both the amp and loudspeakers. In contrast, many vented direct radiating systems have a much lower overall impedance, and the load seen by the amplifier is far more difficult to drive. Any amplifier with a marginal protection circuit may cause spikes on the audio waveform - often well before clipping.
Figure 8 - Impedance Curve of W-Bin
The graph above shows the impedance curve of a 2 x 15" Etone (an Australian Speaker manufacturer) W-bin, measured some 20 years ago (it's been converted from a hand-drawn image). While the nominal impedance is 4 ohms, the actual impedance is at least 6 ohms for the normal frequency range of these boxes, and over 8 ohms for the area where a large proportion of the energy is needed. While users may have thought they were using a 500W amp (for example), in reality the power would have been considerably less than 250W peak, with an average of perhaps 80W or so.
At the time of writing this article, I had a powered sub with two satellite boxes at home (long since sold as of 2016)). The sub used an 18" driver, with a 900W amplifier. The satellites each had their own 300W amp. The system was pretty loud and sounded quite good, but I know from past experience that the same drivers, same total power, but with horn loading throughout would have been much, much louder and would sound better. The compression drivers and HF horns would need a serious upgrade though, or they wouldn't match the midrange and bass. Including the better directivity of the horns, I'd guess at another 10dB with horn loading. To get the sub-satellite system to the same SPL would therefore have needed another 10dB of power - from 1,500W to 15kW, which would (of course) simply blow the speaker drivers almost instantly. That's a seriously big difference. I do admit (however reluctantly ) that the horn loaded system would not fit into the back of a family station wagon or even a large SUV, but the difference in efficiency is astonishing. I've used many horn loaded systems in large venues with a lot less than 1,500W, but the sub-satellite system would only ever be suitable for small pub bands.
The sub-satellite system had some interesting specifications. Maximum SPL for the sub at 1m was claimed to be 137.81dB at 10% THD, and the speaker driver is rated at 101dB/W/m. 900W is 29.5dB above 1W, but strangely, 101dB + 29.5dB is only 130.5dB - almost 7dB shy of the claim. If the 900W amp were pushed to full clipping (producing a squarewave output), it gains another 3dB, but that's just a tad more than 10% distortion (43.5% in fact). Where did the extra SPL come from? It can only be magic, because it certainly can't be explained with maths or science . Strangely, the satellite boxes were rated correctly, although there was no compensation for power compression.
Some major tour suppliers have even devised cardoid (directional) subs - something that a decent sized array of horn loaded subs did automatically. To cancel the sound from the rear of the box, additional drivers are mounted and driven in anti-phase from the main speakers. This makes the overall system even less efficient, because the power fed to the rear speakers is completely wasted - it contributes nothing to the SPL in front of the box, but only cancels the bass as heard from the rear. While very clever and undoubtedly scientific, the power needed to achieve a realistic SPL in a large venue is simply staggering. One I looked at claims 2,250W for a single subwoofer box. A decent sized venue might need somewhere between 4 and 10 of them, so would have between 9kW and 22.5kW of power - just for the subs.
Perhaps surprisingly, there are still a few manufacturers of horn loaded systems - including bass bins. Some small operators have designed and built their own, and a (small) few concert PA suppliers also have high efficiency horn loaded systems, not just for the bottom end, but also for midrange. The top end is still almost exclusively horn loaded, with horns and compression drivers available from many suppliers.
Unfortunately, it's impossible for any major manufacturer to rely on their horn loaded systems to make worthwhile profits, so line arrays form a large part of their offerings. We can also expect to see many more bandpass subwoofers being used. There are already quite a few available, and while these can be extremely efficient, there is something of an art to designing them properly. A high pass filter is essential with any bandpass or normal vented box, because the cone will be completely unloaded at very low frequencies. Bandpass subs (like vented enclosures) can also present a difficult load for the amplifier, so an amp that has well designed protection circuits is essential.
There are several on-line sellers of speaker box plans, with a large proportion of those being horn loaded. I don't know how well the various designs work, but I would expect fairly respectable performance and much higher efficiencies than plastic 'stick-boxes' or line array systems. The usage of these systems is unknown, but I'd expect them to be popular with home builders and budding musicians. They are certainly not mainstream, and it's unlikely that fully horn loaded speakers will ever return to their former glory. I'd like to be proved wrong of course, but that's unlikely.
One new 'trend' that is extremely unwelcome is the proliferation of switchmode power supplies (SMPS), Class-D amplifiers, SMD components and custom ICs that cannot be replaced by anything that one can buy. It is sometimes possible to make repairs where the fault is a failed output MOSFET or some other part that uses through-hole mounting to the board, but there is a great deal of equipment where repairs are simply not possible. This can be due to SMD part failure, and that is often accompanied by wholesale destruction of the PCB, including tracks that are literally blown off the board.
This isn't helped when a complete 2kW/ channel power amplifier (for example) uses one single (large) PCB for everything - the power supply, power amps, input circuitry and/ or DSP (digital signal processing). Even if a replacement PCB is available, the only thing that gets re-used is the chassis and (maybe) the connectors. Better than nothing, but once the supply of spare boards runs out, the entire amp is scrap.
It used to be expected that instruments, amplifiers and PA systems could be repaired if something failed, but we are now seeing a great deal of gear that simply cannot be fixed when it fails. You might be able to get a replacement PCB if the gear is less than 5 years old, but otherwise it's likely that the entire unit will have to be scrapped. This is made even harder when manufacturers flatly refuse to provide service information (some will will even threaten prosecution if you dismantle the product). This is an untenable situation, and causes vast amounts of 'e-waste'. Powered speaker boxes aren't immune - if the amp fails and can't be repaired or replaced, as often as not the whole system becomes junk.
Horns work. Simple as that. Yes, they are large and hard to move around, but in terms of 'bang for the buck' and reliability, nothing else comes close. Because of the horn loading, speaker cone excursion is minimised, so extreme XMAX drivers are not needed. Cooling is better because the voicecoil remains in the gap, and because much less power is needed, there's not as much heat to get rid of. There is still the issue of frequency response lobing when more than one horn is used side-by-side, but even that problem is easily solved, and total power requirements can be lower again.
The Grateful Dead did it years ago with their 'wall of sound' system ... each set of speakers is effectively an independent line array PA system (but not the same kind of line array that is used now). With a completely separate PA for each instrument there is almost zero interaction, and while there is some lobing from each system, it's spread out across multiple PA systems and is far less objectionable. One PA was used for the vocals, another for the drum kit, another for lead guitar, one for rhythm guitar, one for keyboards, etc. By separating each instrument, the overall mix and balance is easily changed, and outrageous SPL can be achieved with relatively modest power amplifiers. See The Wall Of Sound for the history and photos of this system. It is most regrettable that no-one has utilised this concept since, as it is a technique that could make a lot of current systems sound a great deal better than they do now .
Just as biamping a system can achieve close to 4 times the apparent amp power (see Biamping - Not Quite Magic (But Close) for more), splitting the PA does the same, but better. The drum PA can be optimised for drums, the vocal and guitar PAs don't need any subs, the keyboard PA can share its subs with bass guitar - the possibilities are endless. All too easy with the mixers that are available now, but it has always been possible. Unfortunately, the Grateful Dead was the only band to make full use of this arrangement to my knowledge, and they did it mainly from necessity - for the most part, big PA systems just didn't exist at the time.
If a single large PA system runs out of amp headroom and clips, everything is distorted. If separate PAs are used, if one distorts it may not even be noticed unless the distortion is gross and long term. The odd transient that gets clipped isn't audible, but when the entire band depends on a single PA system, then you will need plenty of headroom. With low efficiency direct radiating speakers instead of horns, speaker damage is inevitable unless everything is carefully monitored at all times. This tends not to happen, except at major concerts where the added cost can be justified. Just for the record - line arrays do not (and cannot) address this. They are comparatively inefficient, but are designed to (hopefully) survive the insane power that people expect to pump into them. I don't see this as progress!
Of course, one needs to look at the SPL that's actually required. While it's not uncommon for systems to register a fairly consistent 110dB SPL in typical venues, one must ask if this is really necessary. At 110dB, the recommended exposure time is around 2 minutes in any 24 hour period, after which permanent hearing damage is probable. Even at a rather subdued 100dB SPL, the limit is around 15 minutes! I'm not suggesting that PA systems be run at 90dB - part of the experience of a concert is the volume level and the feel of the bass. To some extent, we (unfortunately) must accept that some hearing loss is almost inevitable, but the excitement factor is easily created without running the PA flat out all night.
One of the tricks I used to use when mixing live sound was to turn the master faders down when the band played quietly. The quieter the playing, the lower the faders ... people would actually stop trying to talk and listen! Since I made it my business to know the music, I knew exactly when the crescendo was due. The faders were snapped up to (almost) the maximum, and a very common comment heard from the punters was "That's the loudest f...ing PA I've ever heard !!". It wasn't (the whole system was about 1,200W), but by having dynamics it sounded as if it was much more powerful. It also adds greatly to the music ... loud bits and soft bits are as essential to the sound as the use of different notes. No-one would want to listen to a band that played and sang at only one note for the entire night, so why should people have to listen to the same volume for the entire gig?
In the late 1960s and early '70s, the mixers used were usually incredibly primitive. A typical mixer may have had 8 channels, all rotary pots (including the faders), pretty bare-bones EQ, and little if anything by way of channel inserts or effects sends (no-one needed effects sends because there were few effects units available), other than a tape echo and (maybe) a graphic equaliser a little later on. The mix was commonly done from the side of the stage, and foldback was unheard of except for a very few larger systems. Once the need became apparent, large format mixers were made by major manufacturers, various sound companies and individuals. 24 channels were usually enough, and most bands got perfectly good results with 12 or 16 channels. Effects racks started to develop once it was apparent that this new 'fad' wasn't going away and effects units became available and affordable. By the mid 70s, one would expect to find an active crossover, a compressor/limiter, and a tape echo machine along with a few other semi-random effects units. Many bands systems also included domestic equipment - especially things like audio cassette players.
Mixers also became much more capable. By the early 80s, mixers were readily available that were not much different from what we see today. The old style 'PA head' that was used with its column speakers was reinvented as the 'powered mixer' in the late 70s. Where the column amp might have had 4 channels with bass, treble and volume (but little else other than a master volume), the powered mixer was usually a reasonably competent mixer with all the expected frills, that just happened to have a stereo power amp built in. These are (still) mainly used for smaller venues, because it has been difficult to get much power from amps that would fit into the available space. Now that Class-D (switching) amplifiers are becoming more common, far more power can be packed into a small space than was possible before.
While I'm sure that there is a great deal more info available than I've got here, it's largely academic. While there have been great strides in technology, the humble analogue mixer was already very good by around 1980, and subsequent additions have just provided more functionality (especially effects in later mixers) rather than make any quantum leaps in sound quality or performance. Analogue circuitry has really only made a few baby-steps in the last 20 years, and most of the improvements are close to the limits of audibility. At over 100dB SPL, there is no audible difference at all. Some of the early mixing consoles are actually sought after today for their 'sound', especially things like old Neve and SSL mixers - some of which almost have a cult following.
Of course, we now have digital mixing desks. These can make life a lot easier once set up, because they offer full automation. The usefulness of automation depends on the musicians, the programme material and the skill of the operator. Regardless of claims though, don't expect the sound quality to be any better than a decent analogue desk. One of the things you do get is the far more flexible signal routing capabilities. This can make it a simple matter to split the various sources into separate PA groups, eliminating many of the problems of having everything handled by one big system. Unfortunately, I don't know of anyone who's doing that. Pity, because that's one area where huge gains can be made, and the final mix can be cleaner and more dynamic (and with less power) than is generally possible otherwise.
As noted above, I either designed or helped design and build PA systems, guitar amps, bass amps and the like. A few of the projects from the 1970s are shown here, along with some info about each. These designs were all in production at the time, and some were used as the basis for a successful hire business. Unfortunately, there are no longer any photos of what I consider to be one of the best PA systems available at the time. I designed and built it in the early 1970s, and used it with a number of different bands. It used particularly high efficiency speakers, was 2-way horn-loaded, and managed to blitz every other system it was ever set up beside. There are many things I'd do differently now, but that's always the way (commonly known as '20-20 hindsight' ).
|Concert PA. Developed in the 1970s for the concert entertainment hire industry. Fitted with 2 x 12in speakers, slot radiators, and internally powered with solid state 400 Watt amplifiers. The moulded fibreglass enclosures could be considered a precursor of today's smaller line-arrays, but they were generally used in groups of no more than 3 midrange and 4 folded horn subs a side.|
|This 16 channel mixer was ahead of its time back then. Each channel had independent EQ, compressor/limiter, LED meter, etc. With 50 metre multicore and 16 channel stage mixer, it even included built in talk-back. Stereo output sends were designed for 4 way active systems.|
|The 8 channel stage mixer was robust and very versatile. Made in an aluminium extrusion, it was designed for the hire industry. It could be external, battery, or phantom powered. Each channel had basic EQ (bass and treble), plus a main and auxiliary send.|
|These moulded fibreglass cabinets were virtually indestructible and specifically designed for the entertainment hire industry. The cabinets could be fitted with a variation of 15in speakers and horns and internally self powered with 200 Watt amplifiers. They were dubbed the 'washing machine' boxes because of their size and because they were white.|
|100W per channel ultra linear valve amp. This beautiful audiophile valve amp was produced in limited quantity for AMW (a high-end speaker manufacturer). The open extrusion construction is strong and unique, and allows for free air flow ventilation.|
There were several other systems we made at the time as well, but photos seem to be long gone. At the time, we operated under the name 'Burnett-Elliott', being John Burnett (Lenard amps) and myself. I toured with a band using one of the concert PA systems, and the combination of sound quality and great music went down well everywhere. Like all systems, the concert PA had some interesting quirks, but they were relatively benign, and the mixer had more than enough equalisation available to iron out the wrinkles.
I'm not silly enough to try to predict what will be next. There are a few people in professional sound who (like me) dislike line arrays and hanker for the PA systems of old, but with a bit more applied science. However, it is very doubtful that we'll see a resurgence of horn-loaded speakers, simply because of their size and weight. There are a few around - new designs with fully horn-loaded drivers still exist and are being manufactured, but these seem to be limited to midrange and the top end. Other than the products from a few experimenters, there are few horn loaded bass cabinets any more. Ample power and bass drivers with huge excursions mean that bass can be delivered by much smaller cabinets than before, but with the ever-present risk of driver failure. This is unlikely to change.
There is a growing trend to use microprocessors (or at least microcontrollers), DSP based systems, and surface-mount components in audio equipment, and these systems are generally impossible to service by traditional means. If a circuit board develops a fault, then the entire board is replaced, and when spare boards are no longer available, you throw the equipment away. This is already happening, but expect it to get worse. While this is a little off-topic I suppose, it is an important consideration - especially for pro-audio gear that's expected to last a long time. In addition, Class-D (PWM) amplifiers are now becoming mainstream. These are capable of extraordinary power outputs with very little heat - the limiting factor is the mains outlets!
Since it's unlikely that buyers will start selecting speaker drivers based primarily on efficiency, we can expect that ever more powerful amplifiers will be unleashed on the poor unsuspecting loudspeakers in systems, loudspeaker manufacturers will desperately try to satisfy the buyers' lust for power (most buyers will continue to ignore efficiency just as they do now), and we'll see more of the same for some time.
We already have loudspeakers that are 'protected' by means of internal series filament lamps [ 5 ], and these can provide us with at least 10dB of power compression - perhaps more. The punters are happy though, because "this 160mm speaker can handle 175W". Few seem to have noticed that after around 25W, it doesn't get any louder, but if they looked inside they'd see the light .
One thing I'd really like to do is take the limiters off some systems, and jam them up the sound engineer's backside. Often, everything is compressed to within an inch of its life, so a solo acoustic guitar is just as loud as the band at full tilt. NO! Music is not like that. It has (or should have) loud bits, soft bits and everything in between. The same is done with CDs and broadcast FM (forget DAB - that's often worse than MP3). Compression is even worse when the system is still driven into distortion!
It's difficult to make any absolute conclusions with such a disparate range of topics, but there are some things that are very obvious. One of these is the myth of power handling and the general inattention paid to cone excursion. These two have seen the demise of countless loudspeaker drivers over the years, and will undoubtedly continue to do so. At the very least, all tuned boxes and horn systems require the use of a high pass filter to remove programme content below the lowest frequency that can be handled by the loudspeaker/enclosure combination. Where amplifier 'headroom' is provided (by using bigger amps than needed), even greater attention must be paid to ensuring that voicecoil dissipation and cone excursion are kept within safe limits at all times.
Using peak limiting is perfectly alright, provided the limiters are set up to maintain at least some dynamic range. This means a fast attack and relatively slow decay - preferably a few seconds if possible. This maintains an acceptable peak-to-average ratio, makes the music sound more alive, and gives loudspeaker drivers some hope of long-term survival.
Issues like lobing will forever be a problem with high power sound systems. Since there is no way to generate the sound power needed with single drivers, multiple drivers are simply a fact of life. With multiple drivers comes lobing (no extra charge ). The effects can never be eliminated, but they can be minimised by careful speaker placement, or by splitting the system so parts of it are used for separate sections (e.g. instruments and vocals).
High distortion is easily produced in the throat of a horn with a compression driver. There is only one answer to this, and that's to keep the power levels low, and use multiple drivers and horns to achieve the required result. It is also necessary to select the optimum system based on your needs, and this can involve a great deal of research. So much of the data you find is either erroneous or simply leaves out the very information you need to make an informed choice. Without knowledge, you are at the mercy of every snake-oil merchant in the business.
It's important for anyone choosing a system to avoid deciding on something based on its (apparent) popularity elsewhere. Elsewhere does not have the same venues that you do, and apparent popularity is just that - apparent. Anyone can write glowing testimonials and place them on their website. Unless you can speak to the actual people who wrote the testimonials, they are meaningless. Also, be wary of people who post in newsgroups and forum sites. While they often seem to be unbiased, you'll find that some have a vested interest in a particular brand, but may 'forget' to disclose this.
It's undoubtedly been noticed that I have a preference for the highest possible efficiency in a system. I know that power is cheap, and that there are drivers that seem to be able to take the claimed power. This doesn't change the fact that power compression is a very real and easily demonstrated problem. Only by keeping the power as low as practicable can you avoid the worst effects of power compression, and the side-issues that are created when drivers (and the air inside the cabinet) are allowed to become hot.
Needless to say, I don't recommend that any high power system be run with passive crossovers. Apart from the fact that they introduce their own losses, passive crossovers also mean that once the amp clips, the entire audio spectrum is contaminated. The ability to manage the signal level in each frequency band can only be achieved sensibly when active crossovers are used, and this gives the skilled operator a system that is louder and cleaner than will ever be possible with passive crossovers - with the same total amplifier power ratings. When passive crossovers are used, you need a lot of extra headroom because of the full bandwidth signal, but you must then restrict the average power to suit the speaker power ratings.
Yes, active crossovers require more amps and possibly cables, but that's why you can get 4-pole Speakon connectors almost anywhere. Remember that horn compression drivers don't need (and can't use) anything above perhaps 100W (allowing for headroom), so amp power requirements are minimal. The small extra bother is well worth the improvement in sound quality.
This is very hard. There were countless sites that I looked at, and while a few had some useful information, many had virtually nothing that was even close to reality. While it would be nice to have been able to put together the history of PA systems, there is remarkably little info available with factual material.
So, the WWW as a whole may be considered the secondary reference, with the rest coming from accumulated knowledge, memory and the few links shown below. There are obvious references to JBL, Altec (Lansing), RCA, Cerwin-Vega and other manufacturers, and some of the photos are adapted from their websites or other sources. Any claim of breach of copyright cannot be entertained, since I only used photos that are effectively in the public domain, as they are published on many different websites.
There is some additional information about horns on the Lenard Audio site, and a lot of additional info about PA systems and the like.
My thanks to Phil Allison, Les Acres and John Burnett for proof reading, suggestions and additional information.
|Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2009. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro- mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference. Commercial use is prohibited without express written authorisation from Rod Elliott.|