BBC-style thin-wall cabinets. Why so special?

[A.S.]

... i still do not quite fully comprehend this lossy cabinet philosophy & how it can be tuned to the advantage of the critical midband... One thing for sure however is that there is totally no trace of cabinet colouration in the midband. How a vibrating cabinet that can have no impact on the critical midband really baffles me...

What underpins the BBC's thin-wall cabinet philosophy (and I was surprised to read that exact word in one of Harwood's papers recently) is the observation that a perfectly cast bell will ring on for many seconds. Conversely, a bell with a hairline crack will sound leaden and hardly ring at all. It's the same with cabinets: if the panels are all rigidly glued together then at some critical frequency or other a note or notes in the music will trigger the cabinet's natural structural resonance. In such a rigid structure, there is nothing that can be done to suppress the ringing - and each time that note reappears, it tops up the ringing which then becomes a permanent drone underneath the music.

Conversely, in a thin-wall cabinet, the lossy joints (i.e. removable baffle/back and the generally 9-12mm thin panels used throughout the box) each act as an acoustic hairline crack. They inhibit the build-up of resonance. Simple as that really!

Now, let's not kid ourself that it is possible to kill cabinet resonance stone dead. It isn't. Not with any approach to cabinet design because the sound pressure inside the cabinet is huge. What the thin-wall approach does is to move unwanted resonances downwards in amplitude and frequency so that they are adequately buried below the music and then pushed down in pitch. Note that I said adequately. Providing that the resonance, be it from the cone, cabinet or even recording - whatever the source - is x dBs below the fundamental, the BBC proved that it was completely inaudible. Once inaudible to trained listeners on all types of music/speech, that is the end of the matter. Inaudible to the trained listener is as good as the solution needs to be. It is neither necessary nor cost effective (nor good engineering) to continue pushing for a degree of theoretical excellence that nobody can appreciate but everyone must pay for. That pragmatism keeps our speaker affordable - and sounding natural.

What we seem to be lacking in the industry today is the good old fashioned common sense that was abundant when serious researchers with zero commercial interest (i.e. the BBC) had their hands on the tiller. Thank goodness that they thoroughly documented their efforts for posterity since physics, acoustics and our hearing are the same now as fifty years ago. Now it seems we are all conditioned by marketeers to chase theoretical perfection which is far, far beyond what our ears can reliably resolve.

[Gan CK]

What underpins the BBC's thin-wall cabinet philosophy (and I was surprised to read that exact word in one of Harwood's papers recently) is the observation that a perfectly cast bell will ring on for many seconds. Conversely, a bell with a hairline crack will sound leaden and hardly ring at all. It's the same with cabinets: if the panels are all rigidly glued together then at some critical frequency or other a note or notes in the music will trigger the cabinet's natural structural resonance. In such a rigid structure, there is nothing that can be done to suppress the ringing - and each time that note reappears, it tops up the ringing which then becomes a permanent drone underneath the music.

Conversely, in a thin-wall cabinet, the lossy joints (i.e. removable baffle/back and the generally 9-12mm thin panels used throughout the box) each act as an acoustic hairline crack. They inhibit the build-up of resonance. Simple as that really!

Now, let's not kid ourself that it is possible to kill cabinet resonance stone dead. It isn't. Not with any approach to cabinet design because the sound pressure inside the cabinet is huge. What the thin-wall approach does is to move unwanted resonances downwards in amplitude and frequency so that they are adequately buried below the music and then pushed down in pitch. Note that I said adequately. Providing that the resonance, be it from the cone, cabinet or even recording - whatever the source - is x dBs below the fundamental, the BBC proved that it was completely inaudible. Once inaudible to trained listeners on all types of music/speech, that is the end of the matter. Inaudible to the trained listener is as good as the solution needs to be. It is neither necessary nor cost effective (nor good engineering) to continue pushing for a degree of theoretical excellence that nobody can appreciate but everyone must pay for. That pragmatism keeps our speaker affordable - and sounding natural.

What we seem to be lacking in the industry today is the good old fashioned common sense that was abundant when serious researchers with zero commercial interest (i.e. the BBC) had their hands on the tiller. Thank goodness that they thoroughly documented their efforts for posterity since physics, acoustics and our hearing are the same now as fifty years ago. Now it seems we are all conditioned by marketeers to chase theoretical perfection which is far, far beyond what our ears can reliably resolve.

Thks once again for the prompt reply & a very good illustration indeed Alan. The thin wall cabinet really does work wonders as far as accurate timbre & musicality is concerned. The importance of the BBC findings & the hard work of all the engineers involved back then cannot be denied & taken for granted indeed. I just finished listening to a set of CDs & LPs over the last 2 hours & am taking a short break to read this reply. After which i will continue to listen till the wee hours in the morning. Its really very addictive listening to music played on Harbeth. Ok time to carry on with my musical journey with my wonderful SHL-5s.

[A.S.]

The reason that the BBC began to investigate alternatives to the thick wall enclosure used in the massive floor standing LSU/10 was simple: the increase in outside broadcasts in the late 1950's (an era when the TV service rapidly expanded) needed smaller, more portable, lighter speakers that could be taken around in small OB trucks, the LS3/1 being the first. For this reason alone - nothing to do with technical research, the thin-wall panels philosophy (to use Harwood's own word) was born. It was subsequently discovered that thin-wall had excellent acoustic properties and this was proven by detailed, documented measurement by the BBC and others (Barlow, Stevens). All classic BBC monitors since the LS3/1 in 1959 have used thin-wall cabinets - the exception I can think of was the LS5/12A.

The attached image (source: Barlow) shows a speaker's frequency response (red trace) hovering along the 40dB sound pressure line - an arbitrary loudness. Also plotted on this graph are three other curves - I've coloured one black, and the other blue, but there is a dashed curve in between them. These three show what a microphone picks-up as the output from the cabinets side, top and back panels alone completely ignoring the useful sound output from the drive unit. You can see that in the middle frequencies, this 18mm cabinet has a peak output around 500Hz which is very nearly as loud as the drive units output! In other words, although the walls appear to be thick and rigid to the eye, to the sound waves inside the box they are as acoustically transparent at this frequency as a sheet of paper. Shocking.

At the other extreme, the box made of 6mm has its peak output well below 100Hz, somewhere around the port resonant (tuned) frequency, which diverts the cabinet's acoustic output well away from the all-critical midband down to bass frequencies. Note however, that the 6mm cabinet has a peak quite close to that of the 18mm somewhere about 500Hz but it is at a lower level. This is not ideal.

Now the clever bit! If you look closely at the 12mm curve (the dashed line) you will see that it has only one peak - at about 100Hz. The peaks at the middle frequencies evidenced on the 6mm and 18mm panels are much suppressed with the 12mm panels. This 12mm option looks promising. Now, suppose we could tame that 100Hz peak and reduce it's level ... and maybe pull it down in frequency a little we'd have a great sounding, resonant free, clean-midband cabinet..

Continued ....

>

[A.S.]

I illustrated in the previous post how the speaker cabinet's panel thickness had a great influence on its resonant behaviour. I showed that despite the visual impression that 19mm panels are rigid and solid, acoustically they are as transparent as tissue paper at a certain problematic frequency. I also illustrated that the problem frequency is related to the thickness of the panel, and that as the panel's thickness increased, the problem frequency moved upwards in frequency encroaching into the midband.

The solution that the BBC researchers described involved using a relatively thin panel - say, 9mm, half the conventional thickness for a speaker box, but loading the inner walls with a rubbery bitumatic material such as roofing felt or its industrial alternative. Being rubbery, this substrate panel flexes with the wood panel, and it's flexing causes the sound waves to be converted to heat in the rubber molecules. Once they are safely locked-up as heat, their energy is dissipated and will not cause us any sonic problems. Attached is an example of a (not very good speaker, but that doesn't matter at all for this illustration) with its frequency response in red. The cabinet made from 9mm panels. The black curves shows the output from the raw cabinet excluding and isolating that from the drive unit and the green curve, the output when the panels have had a layer of bitumen attached. Three things can be observed:

1. The peak frequency where the cabinet is acoustically transparent is about 120Hz. This has been slightly reduced to about 95Hz by the addition of the bitumen counter-layer.

2. The magnitude of the peak noted in (1) has been reduced by about 10dB (by two thirds) by the addition of the bitumen layer to the raw 9mm panel.

3. In the critical middle frequencies (marked in yellow) the 9mm panel + bitumen is remarkably inert.

Conclusion:

Despite ones preconceptions that 'thick panels must be better' in the critical middle frequencies, it is clear that the opposite is in fact true: 'thin is much better' - thin panels are a better solution to controlling panel resonances, and making a speaker box that doesn't sound boxy. Thin can be tuned - thick can't.

I am aware of one manufacturer of tower speakers who recently decided to remove the bitumen lining from his cabinet and replace it by using even thicker MDF walls! I can not understand the acoustic logic behind this move. I can understand the cost saving.

>

[Gan CK]

I illustrated in the previous post how the speaker cabinet's panel thickness had a great influence on its resonant behaviour. I showed that despite the visual impression that 19mm panels are rigid and solid, acoustically they are as transparent as tissue paper at a certain problematic frequency. I also illustrated that the problem frequency is related to the thickness of the panel, and that as the panel's thickness increased, the problem frequency moved upwards in frequency encroaching into the midband.

The solution that the BBC researchers described involved using a relatively thin panel - say, 9mm, half the conventional thickness for a speaker box, but loading the inner walls with a rubbery bitumatic material such as roofing felt or its industrial alternative. Being rubbery, this substrate panel flexes with the wood panel, and it's flexing causes the sound waves to be converted to heat in the rubber molecules. Once they are safely locked-up as heat, their energy is dissipated and will not cause us any sonic problems. Attached is an example of a (not very good speaker, but that doesn't matter at all for this illustration) with its frequency response in red. The cabinet made from 9mm panels. The black curves shows the output from the raw cabinet excluding and isolating that from the drive unit and the green curve, the output when the panels have had a layer of bitumen attached. Three things can be observed:

1. The peak frequency where the cabinet is acoustically transparent is about 120Hz. This has been slightly reduced to about 95Hz by the addition of the bitumen counter-layer.

2. The magnitude of the peak noted in (1) has been reduced by about 10dB (by two thirds) by the addition of the bitumen layer to the raw 9mm panel.

3. In the critical middle frequencies (marked in yellow) the 9mm panel + bitumen is remarkably inert.

Conclusion:

Despite ones preconceptions that 'thick panels must be better' in the critical middle frequencies, it is clear that the opposite is in fact true: 'thin is much better' - thin panels are a better solution to controlling panel resonances, and making a speaker box that doesn't sound boxy. Thin can be tuned - thick can't.

I am aware of one manufacturer of tower speakers who recently decided to remove the bitumen lining from his cabinet and replace it by using even thicker MDF walls! I can not understand the acoustic logic behind this move. I can understand the cost saving.

>

Thks Alan for the above illustration. Its indeed enlightening. To us layman, a thicker cabinet would naturally confer more solid engineering & better build quality. Never knew that a thick cabinet would have such a detrimental effect on the resultant sound. Well at least now i know that the thin wall cabinet has made a very important contribution to the excellent & accurate tonality/timbre aspects that are hallmarks of Harbeth loudspeakers. Of course the contribution from the Radial driver is also crucial to the overall sound of Harbeths. By the way Alan, is the thickness of bitumen critical? Besides lining the internal walls with substantial acoustic foam, have you tried alternative materials like rock wool?

Oh before i forget, what are the sonic advantages of having the bass or midbass driver mounted from inside of the cabinet? (for eg the M40.1)

[A.S.]

Yes, it is counter-intuitive that thin walled cabinet can have lower output in the all-critical middle frequencies than thick, rigid walled cabinets. Of course, this is not news to anyone who has made a stringed instrument: light, thin walls, carefully secured around the edges and with just the right amount of damping here and there make for a wonderful open sound.

I have now combined into one graph the previous two. The red trace is the frequency response of the speaker. Don't worry that it's not smooth - it's just whatever was to hand. The interesting curves are the blue one, showing the cabinet output alone (ignoring the output of the bass unit) when the box is made from 18mm (3/4 inch) undamped panels. Note again that in the middle frequencies the output of the box is very nearly as strong as from the woofer. That implies that the cabinet is acoustically transparent and the sound normally constrained inside the box is setting the 18mm panels into sympathetic motion and they are acting as a huge radiating surface.

Contrast that horrible situation with the box made from 9mm board (half the thickness) loaded on the inside of the panels with a counter-layer of bitumen. I've shaded in white the reduction in box output - down from about 38db to about 12dB, roughly 20 times lower output than from the walls of the rigid box!

But .... bitumen is a by-product of oil, and the price has risen dramatically in recent years. Our cabinet makers complain that we are one of the very few using it these days on cost grounds.

[Labarum]

Alan,

Do the advantages of thin walled design diminish with the cabinet size?

I am supposing that a small panel cannot as easily be damped for a low resonance frequency; and I am supposing that a small panel has less scope to add colourations.

But the whole matter is so counter-intuitive . . .

Your views?

The facts?