Loudspeaker Specifications

Or... "How Did My 100W Amp Destroy Your 400W Speaker?"

by Stephen Court

A t the announcement of my children's birth, the first thing everyone asked was "what did they weigh?" It always struck me as somewhat peculiar.

Of all the personal qualities you could discuss, such as the color of the hair, or how healthy they are, weight seems a strange specification. Fancy telling somebody that you have just purchased a new car "Oh really, what does it weigh?"

Weird & Wonderful
Talking of weird and wonderful specifications brings one to the subject of speaker specs and how meaningful they are - or not - as the case may be. From a design point of view, the manufacturers' specs of a drive unit are obviously important when you have to select a suitable driver from a whole load of available units.

They don't give you the slightest clue to how they will sound, but the printed spec is a starting point. Having narrowed down to a handful of units whose sensitivity, frequency band and power handling matches your requirements, the only way of assessing their performance is to load them in your box and listen to them.

No amount of technical specifications and trendy advertisements will tell you what the speaker actually sounds like.

This is a highly subjective process, and quite often a driver with an inferior spec such as a lower efficiency, just happens to sound better. One is reminded of the story when a supplier to one of the world's biggest guitar amp manufacturers, had just acquired a new all singing and dancing computer analysis program in their R&D department.

Their biggest sales volume was to that guitar amp maker. Accordingly, they put their driver through the analysis. For the first time, they discovered multiple anomalies in its performance which, until then, had proved impossible to measure. The cone was breaking up at high sound levels causing loads of distortion.

With the aid of their new computer analysis, they produced a near perfect driver, and with great pride, presented it to the guitar amp manufacturer. Following a long silence, they pursued their engineers for some response. "Totally useless!" came back the reply.

The fact is that both those products used the identical ATC drive units did not change the fact that their specifications appeared in manufacturer's printed brochures with two totally different output levels.

Accepted Specs & What They "Mean"
Frequency Response 40-18000Hz is the typical first line. Without any indication of amplitude deviations such as +3dB, it is as meaningless as 0-60mph without the time co-ordinate. The problem is, most speakers deviate from a straight line across the spectrum.

Engineers generally understand this because while the drive unit should have an even response, the final speaker may be 'tweeked' to sound better at the expense of this linearity. With the possible exception of a studio monitor, how a speaker sounds musically is, what the finished product is about.

Because the speaker may have been modified to sound better, or because many of the finer response deviations are a function of the measuring system having no relation to the final sound, these deviations are conveniently omitted, or modified by the marketing department to look more attractive, turning a useful parameter into a meaningless one.

Our common belief was that, specifications aside, if any loudspeaker could faithfully reproduce a full drum kit or orchestra, nothing more could be reasonably expected.

Along came a little man from BBC Research with a scruffy old tape on a plastic spool. He had us all running round the studio trying to find a machine which could play at 7ips.

To our amazement, his tape was a rather noisy recording of somebody talking. That BBC boffin shrugged his head and went home.

I queried him later, and he said "Specifications don't mean a thing to me. I've known the man on that tape for thirty years. If the speaker can't even reproduce his voice, what's the point in playing music?" He was Spensor Hughes who later formed the Spendor Speaker Corporation.

Power Handling
This is probably the second most useless of all specifications. Exactly the same speaker can be legally described as 100W, 200W, 300W or 400W depending on whether it's RMS, continuous sine wave, continuous music or peak program, and heaven knows how many AES watts.

At the end of the day, the number of watts is totally meaningless unless you take into account the amount of power compression that all speakers have to some degree, and which eats up those watts.

Reputable manufacturers like JBL, Fane and a few others for example, tend to quote power handing at Zero compression, so 400W will give twice the output than 200W.

At the end of the day, the number of watts is totally meaningless unless you take into account the amount of power compression that all speakers have to some degree, and which eats up those watts.

One rather large Italian manufacturer quoted the same power handling for one of their 15in drivers ­ 600W, but with a staggering 6dB of compression due to its design. This means their driver gives maximum output at 150W and is not 1 decibel louder at 300W or 600W.

Power handling is basically down to the ability of the voice coil not to self-destruct at high temperatures. This is largely down to the glue & voice coil former (which is the same for most manufacturers). Although the best loudspeakers, after all this technology, are still paper cones and wooden boxes, it is the voice coil that has vastly improved over the years. With modern technology, the engineer's rule of thumb in bass drivers, is 100W per inch of voice coil, so a 100mm voice coil is sensibly rated at 400W.

I recall suggesting to one manufacturer who was getting back a lot of blown drivers, to drill a series of holes in the speaker's back plate to help dissipate the voice coil's heat. This he did with good results.

The bass is 2x15in and the manufacturers spec is 100dB@1W, and we use two of them so that's 103dB at 1 watt (not correct).

We horn load them, so that adds 5dB. Then add the low mid drivers, the mid and HF drivers, and somehow they arrive at a total of 108dB@1W.

Now we rate this system at 800W RMS. Because doubled wattage gives a 3dB increase, then as 1 watt is 108dB, 2W is 111dB, 4W is 114dB and so on. 512W therefore gives 135dBm. With a total power handing of 800W this system therefore has a maximum sound output level of 138dB.

Nonsense! Believe it or not, I have heard all these versions from different manufacturers, and it is the one spec that seems to be under control of those spurious marketing departments.

Impedance
There is not a lot you can say with this spec. Any speaker will vary from a few ohms at just one minimum point, then rise to all sorts of values everywhere else in the spectrum.

The spec should really say, Nominal or Minimum impedance, but don't forget if you drive a speaker hard, it gets hot and the impedance goes up anyway. The only real use of an impedance figure is to make sure you don't go down too low for your amplifier's rating.

The Bottom Line
Read the specs by all means, but don't take them too seriously. You cannot blame the marketing department for modifying the specs because they are paid to sell as many speakers as they can.

It's a lot of money to spend at the whim of a sales brochure or what somebody else has told you. There is only one way to audition a speaker, and that is put it beside another speaker in the same room, using the same power amps and cables and do an A-B test.

...that particular driver displayed a distinctive cone break-up which emphasized the best characteristics of the electric guitar.

Also, have it installed by the same technician because I have seen some wicked tricks with salesmen's demos. Any reputable manufacture would welcome such an opportunity to demonstrate the prowess of his products.

If he doesn't have time for that, or tells you he doesn't need to do that because his products are so popular, you have probably saved yourself a lot of money. No amount of technical specifications and trendy advertisements will tell you what the speaker actually sounds like.

The real qualities of a speaker, like definition and clarity, the bite of a snare and the kick of a bass drum are impossible to describe with words and figures, so go for a demo and have confidence in your own ears. They were, and always will be the best judge.

Stephen Court of Court Acoustics in England pioneered much of the work on high power touring systems in the sixties and seventies, and produced the first music line array - Black Box in 1975.

His work on evaluating speakers led to his development of the test CD Sound Check which he co-produced with Alan Parsons at Abbey Road Studios in London, and which became the best selling test CD used world wide as a reference for the subjective evaluation of speaker systems.

Root Problems
The principal problem in all of this is that the accepted specifications for drive units have simply been applied to complete loudspeakers. Despite the numerous scientific principals common to drive units, the final and singular purpose for them is to convert electrical energy to acoustic energy, a process which can be definitively measured.

However, a complete loudspeaker has a much more ethereal purpose, to convert electrical information into music. This is a highly subjective assessment, and the standard specifications, which may be useful in the selection of drive units, are virtually meaningless in the selection of a loudspeaker system.

In other words, the manufacturers specifications as they stand, have little or no bearing on the sound of their speakers. More to the point, is that without some form of discipline in their issue, they are at the whim of whoever prints the brochures, making them even more meaningless.

We are all aware of speaker reviews in the hi-fi press, with a proliferation of ludicrous terms like 'boxiness' and 'cardboardiness', but how do you express clarity and definition with language? It's a bit like explaining the difference between red and blue to a blind person.

What makes this so unfortunate is that many customers rely on speaker specifications to assist them in the choice and purchase of a loudspeaker system. Consider the recent trade when a customer visited the Martin Audio stand, and commented that he intended to purchase an EAW speaker because it had twice the output.

If one particular manufacturer of stage monitors is to be believed, this may even be accompanied by a frequency response graph which displays an incredibly even response (Think: Utah Salt Flats) before dropping off, not at the standard 12dB per octave dictated by the laws of physics, but at an incredible 47dB per octave.

Our sales director once asked me for more response curves to assist his marketing efforts against our competitors. I explained that I didn't object to showing fellow designers our curves, but if they were to be published as a sales aid, I assured him that would be the end of his sales.

I explained that the pen recorder an engineer uses, is quite different from the marketing department's pen recorder, which has an infinitely high paper speed and an infinitely low pen speed in order to reproduce the nice flat line shown in sales brochures. Again, a useful specification is turned into an absolutely meaningless squiggle.

Another Lesson Learned
Long ago, I was auditioning studio monitors for a major studio in London. We all turned up with our measuring equipment, test tapes and recordings of drum solos, orchestral spectaculars, organ pedal notes and so on.

This innovation would have resulted in extended life for that driver, but those wonderful men in the marketing department, saw the improvement, and instantly upgraded the speaker to 1000W! As a result, the customers just turned the wick up and the manufacturer got even more returns.

Sensitivity
Usually measured as how many dB you get at 1 meter distance, with 1 watt input. This is a useful spec on the face of it, because it shows how efficient the system is, and the more output per watt, means more headroom, less distortion, cheaper amplification and so on.

Nice in theory, but how is it measured? Logically you put a voltmeter across the speaker terminals, and an amp meter in line with the terminals. When you have 1V and 1 amp, you have 1 watt - simple!

To make things easier, some manufacturers, rather than measure it, use the stated impedance (i.e. 8(omega), then apply the correct voltage calculated to produce 1 watt. Instead of the 8(omega) impedance, others use the DC resistance (approximately 6(omega)), so they get a different sensitivity figure.

Some manufacturers don't even bother to do that, they simply take the sensitivity published by the driver manufacturers, add up the number of drivers (which shouldn't make any difference), then add a spurious figure because their driver is horn loaded which produces more signal on axis than the same speaker on a flat baffle.

The result with so many system manufacturers, is a totally false figure, even to the point where two manufacturers show different sensitivities even though they are using exactly the same drive unit.

Maximum Output Level
This is possibly the daftest spec of all. How manufacturers measure this totally escapes all of us. How do they test it? What with?

Do they play a sine wave at a single frequency which gives the most output, then with a sound level meter measure the result when they turn the wick up? To get a maximum figure do they keep on turning it up until the speaker blows up, then measure the level just before it exploded?

Do they play it in a small absorbent room that sucks up the sound, or a large reverberant hall which may offer 20dB of room gain? Do they use pink noise, which most fairly covers the 20hZ.

to 20kHz spectrum?

Do they play music? Which is more real. If so, do they read the average level or the peak level? And then, over which period of time? So it goes on.

Still, the most popular output measurement, because it's the easiest and definitely most inaccurate way, is to calculate the maximum output level.

Having said that, people who connect speakers to give 2(omega) impedance cannot be too serious about sound quality anyway.

Throw
This wonderful spec that is now all the rage. Polar diagrams of individual drivers and complete systems is a useful specification if measured properly and left alone by the sales department's graphic designers, but I suppose that is really too much to ask.

We are now told that "Our bass bin had more throw than his." Wonderful how these salesmen have got round the fundamental laws of physics. We used to talk about dispersion, but now it's all throw.

The fact is, a conventional speaker system has reasonably wide dispersion. If you reduce the speaker's horizontal (or vertical) dispersion, it will be louder in front - it throws more. End of story.