In the first of a new series of posts, Senior Product Manager, Mike Gough, looks at some of the tools used by the professional speaker designer including their limitations and pitfalls.
The anechoic chamber
Anechoic means “no echoes” and, if you have never seen an anechoic chamber, this is one of the best. It’s at the Building Research Establishment, just north of London, UK and is the largest in Europe. It’s size is indicated by the Bowers & Wilkins engineer in the photo and the speaker is an early Model 801, which puts the date somewhere in the 1970s. We used to hire the chamber in those days to make really accurate bass response measurements. The open air might have been an alternative, but there you have to cope with noise from wind and passing traffic and worry whether the sun is shining or there’s snow on the ground.
The longer the wedges the lower the bass
The walls, ceiling and floor are lined with wedges. They’re over a metre long and absorb sound, partly because they’re made from fibreglass, a very good sound absorbing material, and partly because the tapered shape blurs the effect of having a sudden change of material. Having the taper alternately vertical and horizontal also helps. As a general rule, the larger the chamber and the longer the wedges, the lower into the bass you can measure accurately. Because of its size, this chamber needs an open metal grid above the floor wedges to support objects under test and allow people to walk around.
As you may guess, it costs a fair amount to build a chamber like this and you need the space for it. Nowadays, its use is reserved for when you want to test much larger objects than a loudspeaker. For those familiar with London’s Albert Hall, the discs in the roof were tested here before being installed to reduce that hall’s famous echo.
The bass response of a speaker is now more predictable using modern computer modelling techniques, which allows us to use a much smaller chamber. This is the one at our Steyning R&D facility. The outer dimensions are a much more manageable 5.5m x 4.2m x 4.2m high. It’s accurate down to around 200Hz and here it is shown during a simple tweeter measurement. The wedges are still a metre long, but are made from open cell foam. Having once accidentally fallen into fibreglass wedges and suffered a good week of itching as a result, I can testify that foam is a much more acceptable material.
It’s as important to stop sound and vibration getting into the chamber from outside as it is to stop internal reflections. Of course, the wedges absorb sound in both directions, but our chamber is also isolated from vibrations in the building, hence the gap between it and the floor.
Why we need to measure
At this point, it’s worth taking a step back to ask why the speaker engineer needs measurements at all; surely the final arbiter of what is good and bad should be the ear. That is certainly true, especially as we can’t yet measure every property of our equipment that goes to make up the total listening experience. But, however good our ears are at telling us whether what we have is good or bad, they are not very useful at quantifying how good, or pointing us in the right direction to make any improvements. Only when you can put a figure a property can you tell if you are making an improvement. You have to measure.
Now, nobody in their right mind would want to listen to music in an anechoic environment. Listening in a room or auditorium of some type is much more satisfying and that’s what we do, so why not measure under the same conditions that we listen in? Well, let’s look at some basic frequency response measurements. This graph shows three measurements of the same speaker. In each case the microphone is placed directly in front of the speaker.
The green trace is an in-room measurement of the speaker close to a rear wall. The grey trace is with it closer to a corner. Both these measurements are very ragged and it’s difficult to decipher what is going on. They show the effects of reflections and resonance modes in the room and, while they are similar at higher frequencies, they’re very different in the bass. If you were to move to a different room, the results would be different again. Fortunately, the ear is able to cope with some of these differences, but even then, the professional engineer cannot optimise for just one room and even one position in that room. He needs to know what the speaker itself is doing and assign measurement targets that will make it sound good in the majority of practical listening rooms.
The red trace is an anechoic measurement of the speaker. It’s much smoother and, more importantly, it is consistent and repeatable. In the bad old days, it was considered the right thing to make the speaker’s response extend as far as possible into the bass and measure ruler flat as it did so. If nothing else, it was good for published specifications. Actually, that’s about all it was good for. Nowadays, most designers have more sense and make some allowance for the bass boost that virtually all rooms will provide.
What comes next…
We’ve briefly touched on frequency response, but the anechoic chamber is used for a much wider range of measurements that benefit from this type of environment. However, a discussion of distortion, dispersion and so forth belongs in a future article. And before fans of impulse response measurements point out (justifiably) that you can make ‘anechoic’ measurements in a live room, we’ll compare that approach as well.
For a larger selection of anechoic chamber images, please visit our Facebook gallery.
Mike Gough, Senior Product Manager.