Outdoors on a warm, windless day makes a great environment for carefully measuring loudspeakers. This April 2007 has been especially warm, and today is the first time this year that I've been able to set-up outside for a serious measuring session.
Spending an afternoon in an English wood is a wonderful experience especially at this time of year when the bluebells (Hyacinthoides non scripta) are in bloom. I can't imagine what the birds and rabbits made of the test tones - you could almost hear them trying to make sense of it - but I was ever aware that I was invading their space and that they'd be there long, long after that strange man and his noises had gone.
First I measured our BBC reference LS3/5a as I had the original BBC anechoic chamber curves to hand and it's easy to make a cross-comparison because of the distinctive response shape, then the Monitor 40 at about 4m above ground.
A little QuickTime video I made: http://www.harbeth.co.uk/aboutus/ourvillage/images/m40_in_the_woods.mov
and here is a higher resolution image (big file): http://www.harbeth.co.uk/aboutus/ourvillage/images/in-the-woods-xroads-c.jpg (I'm told this makes a great screensaver!)
In a recent hi-fi magazine they commented that there is a picture on our C7ES-3 web page of its development using the local Village Hall as a quasi-anechoic or 'free-field' space. They write that 'other companies use huge anechoic chambers'. I'm not so sure if that really is the general position. I'll explain.
What is an anechoic chamber used for? It is used for testing the speaker in a reflection-free environment, or at least as reflection free as possible. A 'real' anechoic chamber prevents reflections by lining all the walls, ceiling and floor with thick wedges usually made of foam several meters thick. You can imagine that not only is the cost of such a specialised building in the million dollar range but that even with thick wedges it is not possible to soak up all the sound - so, by implication, there will be reflections and the chamber will never be 100% anechoic. This will cause problems especially when measuring the low frequency response of the speaker.
Now this in not a new problem; the measuring environment has been the bane of speaker designers for fifty years. Once, there were several anechoic chambers in the UK but the last I'm aware of (at the BBC Research Dept.) is shortly to close and the site be redeveloped for housing. It, and presumably all previous ones, didn't have enough trade/income to justify their running costs and space. So what's the alternative?
We're trying to measure the speaker and not the [environment + speaker]. We must avoid reflections polluting the frequency response as seen by the microphone. We can tackle this in two ways: using an anechoic chamber where the reflections are minimised (by absorption) or we can make a measurement in a normal room and (theoretically) use the power of the PC to isolate the reflections from the direct sound arriving at the mic since they will appear on the screen as noticeable glitches. (A fancy word for our post-processing is 'time windowing'). We can then combine a measurement made very close to the speaker (for the low frequencies) with a windowed one made further away for the mid/high frequencies and we can create a quasi-anechoic response just as if we were in a huge chamber.
These days in the 'post-anechoic chamber era' the designer would use his PC to make acoustic measurements on his bench in the lab amongst all the other clutter and noise you'd find there. I go to the not inconsiderable bother of actually creating some space around the speaker in the Village Hall or outside in the woods where the reflection glitches are even more dramatic and can be windowed out with confidence. Who'd build a real anechoic chamber now that we have this analytical power in our PC's? No one.
I have been looking back over my archives to see when this near field/far-field stitching technique was first advocated and found mention of it in 1980 by HiFi News' technical expert, James Moir even in the pre-PC days. (Picture attached of his rig - I'd forgotten all about this. A very superior affair at 7m height! Dangerous too!)
The lab-based quasi-anechoic technique is used by most speaker manufacturers these days and is also the way Stereophile magazine (to name just one) makes their speaker frequency response measurement outside their office. Stereophile Magazine measuring set-up: http://forum.stereophile.com/photopost/showphoto.php/photo/135
But beware! The application of this stitch technique is not as easy as it sounds ..... I speak from long personal experience. So, by modern standards, the luxury of hiring the Village Hall or inconvenience of setting up in the woods is actually the height of sophistication and real dedication and can give results equal to or even better than a real anechoic chamber when carefully time windowed. I see that a representative of Dolby Labs gave a paper on this subject at the AES in 2004 - see attachment.
Wikipedia entry for anechoic chamber: http://en.wikipedia.org/wiki/Anechoic_chamber
Background paper on the influence of the environment on speaker measurement (let alone performance): http://www.listeninc.com/files/pdf/a_comparison_of_techniques_for_evaluation_of_louds peaker_performance_at_low_frequencies.pdf
Paper explaining how good measurements can be made in ordinary rooms (with certain caveats) and modern computer based audio test equipment: http://www.listeninc.com/files/pdf/AESFreeField.pdf
So, back to the woods we go .... it really is the best technique but rather inconvenient and needs a warm, windless, dry, quiet day! Needless to say, if you have the luxury of an existing anechoic chamber it's single biggest advantage is 24/7 operatability.