Features

The Fletcher-Munson curves show that both low frequencies and overall listening levels must be relatively loud to hear natural bass; anyone who has conducted practical speaker tests becomes aware of the fact that the ear is very sensitive to changes of even a few dB at low frequencies.

This means that when a speaker is placed in a home listening room, and you are discussing bass below about 150Hz, the "rolloff" point should be measured at an spl of at least 95dB, and at 100dB for a really good high-end system (footnote 2). The tendency to measure bass response at lower signal levels is simply absurd, given the insensitivity of the human ear to bass frequencies and the measured spl of bass notes in live music (footnote 3).

Similarly, a small loudspeaker, or any loudspeaker in which the bass response falls off at a very sharp rate, would have its frequency response limit set at the point where an anechoic or average room test shows a drop of around 1dB. A speaker with a slow falloff curve can have its frequency response set at the -3dB point, but this simply doesn't work with most small loudspeakers. It adds 5-15Hz of bass response they really don't have under real-world conditions, and disguises the point where they begin to run into trouble.

It is interesting to see what happens when these criteria are applied to real-world loudspeakers. Let's be generous: we'll use a 3dB drop, and, to be neutral, use the anechoic tests run in Britain by British reviewer Martin Colloms (as presented in Hi-Fi Choice).

Let's also be generous and accept the fact that some of these tests are run at 86dB because, as Colloms states, "miniature speakers are generally in gross overload at low frequencies"---although I would strongly argue that such testing is fundamentally unrealistic because it disguises the true nature of speaker bass performance at real-world listening levels.

We end up with the following -3dB response points for some well-reviewed small speakers and mini-monitors:

Speaker, Bass Cutoff
AR-18, 70Hz
BBC LS3/5A, 95Hz
Celestion SL6S, 70Hz
Goodmans Maxim, 120Hz
KEF C10, 110Hz
Mordaunt-Short MS100, 95Hz
ProAc Tablette EBT Super, 95Hz (?)

This kind of performance simply is not compatible with the demands of real music, and has not had any justification in design terms for years. Whether or not one likes the acoustic-suspension principle, it has been possible to design decent woofers and enclosures at reasonable prices for more than two decades.

This kind of performance also gets much worse when we start talking about performance at 96-100dB and the resulting impact on distortion. While speaker size is scarcely an indication of low distortion, the fact is that most mini-monitors effectively clip with bass peaks and what are normal listening levels with other designs. This, of course, is a damn good reason to publish measurements for second- and third-order distortion in loudspeakers at every frequency, and for publishing the clipping level, or point at which the drivers really begin to distort. But, then, publishing such measurements would probably put two-thirds of today's speaker manufacturers out of business!

I don't mean to say that deep bass is vital to good home listening---in fact, in most homes it is difficult or impractical. However, I find damn little excuse for speakers costing over $150 that can't go down to at least 60-70Hz at 95dB SPL with reasonably low distortion.

Speaker measurements of bass performance should also examine real-world dynamic ranges, and be measured at levels no lower than 90dB, and normally at 100dB. I am saying that virtually all of today's small "monitor" speakers are measurably incapable of high-fidelity reproduction, and have no place in a decent audio system.

---

Footnote 2: Most rooms do reinforce the bass of a loudspeaker, but a given speaker may have very different bass-response characteristics in different rooms. Speaker manufacturers generally specify anechoic response because there is no standard listening room. Even so, an anechoic chamber runs into increasing measurement problems below 100Hz, due to its finite size.

Footnote 3: Reviews which refer to the fact that bass power exists in speakers with sharp bass rolloffs below the -3dB point are largely absurd. Most small loudspeakers virtually hit a brick wall at this point and will only have audible bass response below the -3dB point if their midrange and treble are played at painful spls. This is one of many reasons why audiophiles play their speakers far too loud, often to the level where the woofer can no longer respond effectively. This kind of bass often has well over ten times more distortion than the midrange, and is generally blurred, with gross overhang. In most cases where level is used to extend apparent bass, the entire speaker system will be driven to the point where distortion extends well into the lower midrange. In addition, room-interaction problems will be sharply increased.---AHC

Footnote 4: Martin Colloms' anechoic tests in Hi-Fi Choice and HFN/RR do include distortion levels at a 96dB level; whether these will be absurdly high or not depends on a number of factors. The Mordaunt-Short MS100, for example, has levels of second- and third-harmonic distortion not much above 1% at this level, due to the designer not trying to push the drive-unit too low in frequency. He chose, at the price point of this speaker, to sacrifice extension in order to get benefits in reduced distortion. By contrast, the smaller, significantly cheaper, MS10 from the same designer has more than 4% low-frequency THD at this level, this being part of the tradeoff necessary to reduce price. If cost is no object, then any good designer can produce loudspeakers with extended bass at high sound levels; cost is always an important factor, however, both to the designer and to his customers, and something must be sacrificed.---JA