Features

Martin Colloms responds to Tony Cordesman's recent dismissal of small loudspeakers, from Stereophile Vol.10 No.5, August 1987

I must take issue with AHC and the results he presented in "Who Stole the Bass?" in Stereophile Vol.10 No.3. While accepting that his outright condemnation of small monitors was presented as a minority report and that the article essentially concerned personal taste, it was nevertheless backed up by a technical argument in support of his opinion. Moreover, he misuses anechoic data from my own reviews to aid his presentation.

It is hard to know exactly where to begin, the article being such a complex mix of half-truths and opinions. My understanding of his position vis-à-vis the importance of low-frequency extension was that he would prefer a 7-litre Cadillac to a fuel-injected Rabbit: virtues such as precise handling, neutral cornering, grip, and low body roll could all go hang in the quest for low-end torque.

In my opinion AHC is wrong about the importance of the low bass. Midrange is where we live, where we speak, and, as often pointed out in these pages by J. Gordon Holt, if that is wrong, it doesn't matter if other areas are perfect. Good low bass is a nice bonus if you can get it, but no amount of low-frequency wallop can make up for a flawed midrange. The raison d'etre for most small, competent monitors is an excellent midrange performance; excellent, that is, in terms of speed, clarity, and focus. In addition, there are ideal places to site speakers in real rooms, and physically small speakers can take advantage of this in a manner often denied the behemoths.

By taking my measured anechoic -3dB responses for tested miniature loudspeakers (footnote 1) but failing to read my accompanying text on the room-integrated curves, or the discussion on room loading and rolloff rate, AHC has misrepresented the data and the resulting opinion.

While it is true that the low-frequency response of rooms is irregular and difficult to define, this is no reason to dismiss the subject and fail to come to terms with room effects. It is a fact that a listening room is not an anechoic chamber---the latter is a place with no reflections, and approximates to free, open space. Given the dimensions of typical rooms, and given a typical box loudspeaker, it is well known that the speaker's output below 250Hz becomes spherical or omnidirectional, analogous to a naked light bulb. At frequencies below 200Hz, a free-space, stand-mounted loudspeaker radiates sound whose wavelength approaches and exceeds the path lengths from the speaker to the near boundaries.

Let us assume a sensible stand and position are designed for the most even distribution of low-frequency output---typically a 0.5m stand height, 1.2m spacing from the side walls, and 0.9m from the rear walls. At frequencies in the 60-80Hz range the floor boundary is coupled, reducing the speaker radiation from 4pi (spherical) to 2pi (hemispherical), resulting in a 3dB lift. At 45-60Hz, the wavelengths are longer and the rear wall is now coupled, giving a radiation angle of pi radians and adding a further 3dB. Ultimately, at still lower frequencies, the side wall couples, adding another 3dB (fig.1).

Fig.1

By the time 30Hz is reached, therefore, the mean bass lift in a real room relative to the midband, averaged through the listening space, is no less than 9dB! This is precisely the reason why loudspeakers with flat measured anechoic responses extending to very low frequencies will sound bass-heavy in all but the largest rooms, and why small speakers of slow rolloff rate and reasonable power handling can generate significant bass power. Take the Celestion SL600: JA will testify that at quite decent 95dBA mean sound levels (near to realistic live orchestral volumes), a correctly sited pair of SL600s will deliver competent levels down to 35Hz. I freely admit that an audible 20Hz is out of the question, but could AHC quote a number of recordings where such a bandwidth is essential? The SL600 does not produce earthshattering or nausea-inducing bass, but such overkill is not relevant to the vast majority of records. It can and does produce quite reasonable bass levels at realistic but not extreme sound levels.

A look at the figures is revealing. While the '600 is -3dB at 65Hz, its rate of rolloff is quite slow. Consequently, it is -6dB at 55Hz and -9dB by 40Hz, a rate of rolloff quite similar to the rate of boost provided by the local boundaries. In-room, it can be shown by competent measurement that the SL600 is typically 3dB down at 30Hz, an octave below the limit implied by AHC.

Conversely, another small monitor of higher sensitivity, this time using bass-reflex loading for the woofer, is maximally flat to 70Hz but rolls off at a high 24dB/octave rate at lower frequencies, eg, by 35Hz it is 24dB down. 9dB of room lift at this frequency cannot do much for it as it is still 13dB down, and such a low level of bass will indeed be inaudible.

Measurement of a number of larger US-designed speakers has shown irresponsible degrees of bass lift, even "boom." Just as the Germans appear to like an unnaturally bright balance, so some Americans appear to favor overpowering bass, as if this represents the true measure of the emotional impact of high fidelity.

Don't get me wrong---I can certainly appreciate the merits of a really good, large, high-power, wide-bandwidth speaker system; but there are so few of them. AHC does have a point on costs---small British speakers are expensive in the US. However, I cannot see the realities of intercontinental shipping costs as constituting a prior case for the indictment of the high-quality miniature loudspeaker.

I have the advantage over AHC, having carried out many blind listening tests on a huge variety of speakers (footnote 2), both large and small, ranging from original early IMF transmission lines to miniatures such as the LS3/5a and the SL6/SL600. I agree that bass extension matters, but only when a number of other parameters have first been satisfied: tonal balance, midrange coloration, treble purity, transparency, transient definition, stereo focus, and depth. Blind listening tests indicate that bass extension seems to rate sixth or seventh on the subjective shopping list. Many, many times, a smaller, cleaner-sounding speaker has been preferred to a larger and more extended design, due to the latter's inferior directivity and focus and higher levels of coloration.

To conclude, the vital measurement for bass response is not the anechoic -3dB point, whatever that is. The correct method is to optimally site the speaker in a well-proportioned room, and take a computed average of 64 or 80 readings over the listening space, height, and lateral spread. Only then will the sound power at low frequencies be properly integrated into the measurement. A measured 25Hz to 5kHz in-room response, ±2dB, is possible with such a method using 1/3-octave analysis, and it can sound that way too!

Footnote 2: To date, some 400 models.