BASIC LOUDSPEAKER THEORY

~ UNDERSTANDING CABINET BEHAVIOUR ~

Some basic principles of conventional cabinet behaviour

~ DRIVE UNIT THEORY ~ coming soon

Beginners guide to understanding loudspeaker drive units

~ SIMPLE CROSSOVER DESIGN ~ coming soon

Basic concepts of crossover design

An introduction

This is not intended to be anything other than a guide to the effects of various aspects of conventional box cabinet construction. It is not description of what is right or wrong, but may reflect my experiences of what has worked for me over the years.

All cabinets resonate. The trick is to minimise this resonance, and remove any that remains quickly and inaudibly.

One of the biggest advances in loudspeaker performance came in the late 1960's in the UK at the BBC (see drive unit theory). The historical BBC approach at this time was built around thin wall, heavily damped plywood panels, with front baffle and back fixed by screws, providing further 'lossy' mechanical coupling. One of the first commercial loudspeakers to use plastics technology for drive unit cones was the Spendor BC1, a landmark in loudspeaker design. This model was constructed of 3/8" (9mm) plywood, battened at the joints, with 3/8" bitumen impregnated felt bonded to the inside of the panels. This provided a low energy storage, due to the thin panels, with low resonance due to the very heavy damping. Although this speaker broke many modern 'rules' it still sounds good today, and many large manufacturers have a pair for reference use. It is a good example of the need to think beyond the obvious to achieve acceptable results. For instance, the cabinet measures 12" x 12" x 24", which theory says will sound very resonant. Indeed, all the resonances are concentrated in one area, and the overall effect is of perhaps a muddy, flabby lower bass, but with outstanding upper bass and midrange performance (the main criteria at the time for voice monitoring purposes).

This thin, damped panel approach is still employed successfully by some companies, but often now with thicker front baffles and internal bracing. Personally I doubt that this approach can ever provide the 'fast' bass response required by those who listen to modern music with it's inherent dynamic rhythm and pace, due to flexing of the panels at LF. The preferred material these days is MDF. I personally feel that 18mm(3/4") is suitable for most applications up to 40 or maybe 50 litres. Above this you may want to use a thicker material, or perhaps employ careful bracing techniques (discussed later). For smaller than about 20 litres, 12mm(1/2") may be acceptable. Whatever is used, it is important that all corner joints (including front and rear panels) are very rigid, using batten along their full length. Note that there are no rules for panel thickness, as there are good commercially available 50 litre speakers using 9mm thick panels, and also 10 litre speakers using 25mm thick panels.

Some basic principles

As I say, all cabinets resonate. The purpose of a cabinet is to support the drive units rigidly within space, and to control output from the rear of the units. Do not misunderstand this. The best mounting is probably a single large flat plane, infinitely stiff and of infinite size. Reality is that the plane is limited in size, meaning that at frequencies below where the size interacts with the wavelength, cancellation occurs resulting in poor bass extension (this is apparent in the open backed electrostatic designs. Sealed boxes are an extrapolation of this 'infinite baffle' configuration, where the plane is folded. Ported enclosures are a modification of this principle, using the well documented helmholz resnance to boost bass in a narrow tuned band. The trouble is that when you 'fold' the plane, a lot of extra complications creep in.

Unfortunately conventional rectangular boxes consist of several plane surfaces, clamped only at their edges, each plane dissipating it's energy in accoustic output with poor decay characteristics. This is the source of box colouration. As stated earlier, thin panels tend to flex excessively at LF, but making the panels thicker just moves the resonant frequency higher. It does not reduce the magnitude, the decay time becomes worse and the thicker, higher mass panel is harder to control (thicker damping pads must be used). This doesn't mean that you shouldn't do it, just that you should be aware of the consequences.

Not only can undamped panels can have delay times of up to 1 second, but the highest output can approach the output of the drive unit itself ! The audible colouration and smearing effects of this are nothing short of catastrophic.

More on panel damping

By now the importance of damping is self evident. So how to do it ?

The most common method used in commercial loudspeakers is the bonding of bituminous pads to the inside of the panel. These are similar to the type available from auto accessory shops for damping car panel resonances, but are usually somewhat thicker (several layers can be bonded together to achieve the desired thickness). Suitable pads may be available from specialist audio shops. If you use this method, then there are several basic rules to follow :-

The pad should cover the central 50-60% of the surface area of the resonant panel (note : if any bracing is employed, this will subdivide the panel into separate areas to be treated as separate panels - see section on bracing below !).

The pad must be bonded to the panel as effectively as possible over it's full surface area (this is also important between pads if more than one layer is used).

As a rough guide, the pad thickness should be half the thickness of the panel thickness (possibly more on large enclosures, less on small enclosures).

This is not the only method, but is probably the simplest. Damping is all about dissipating the resonances through frictional losses, therefore any material with high internal frictional losses may be suitable. As an idea, a liquid bitumen type compound such as car underseal may work, especially if mixed with sand or some type of mineral. This could be sandwiched with another thin panel to keep it in place. Some models have used lead panels secured with rubber mountings, or even concrete. With a little thought there are many possibilities.

Brace yourself !

Cabinet bracing is a subject worthy of very careful thought, the consequences of which are not immediately obvious. With strategic bracing, it may yet be possible to utilise thin wall, low mass panels with their inherent damping and energy storage advantages. I will attempt to guide you on this, but as you will of guessed by now, this whole cabinet construction thing is about trial and error, but with a little common sense you can circumvent any obvious mistakes.

Firstly a short analysis of panel bending modes. The fundamental is where the panel edges are stationary, and maximum displacement is at the centre of the panel. Subsequent modes occur at twice the frequency where the panel is stationary at the top and bottom and centre, and the two halves are displaced in antiphase. This also occurs from side to side as well as a combination of the two. The same occurs with the panel divided into three in vertical, horizontal and combination planes. This carries on with decreasing displacement and importance. From this we can see that braces placed at any significant null points (such as the centre of the panel) will be utterly useless at controlling these modes. Apart from the B&W matrix design which effectively couples all parts of the cabinet to each other via multiple interlocking panels, the most effective solution is to use offset diagonal bracing to control as many modes as possible.

Another method that seems to work well on thick panels is a damping brace. This is easily acheived by fitting a brace coupled at the contact points by a fairly stiff foam or rubber, therefore placing the panels under tension by virtue of the absorbent material.

It seems obvious that cabinets where the panel dimensions are equal or are multiples of each other will give the worst possible results, and everybody will tell you that the 'ideal' golden ratio of  1 : 1.6 : 2.3  is best. Undoubtedly this is a safe bet, but remember the Spendor BC1 (1 : 1 : 2) mentioned earlier?

Note that this is a simplification of what is actually happening, as the combination of bending, standing wave, pressure and air stiffness modes may all coincide in ways that are impossible to predict. As I say, this is just a beginner's guide.

Internal standing waves

Apart from keeping to safe cabinet ratios so as not to emphasise particular frequencies, control of internal standing waves is by absorbent foam or wadding. Heavy stuffing is usually only used in sealed box systems, as in reflex ported systems standing waves are far more audible and therefore must be more carefully designed so as to take account of the main standing waves. Not only will heavy stuffing interfere with the function of the port, but is also likely to move around absorbing bass energy. Well designed reflex systems often use polyurethane foam sheet of up to 50mm thickness to line the inside of the box. A more effective method would be to attach absorbent centrally within the box maybe to any bracing employed.

Standing waves may also be audible through the cone of the driver itself. This may be particularly apparent in midrange units where, although the cone area is small, the enclosure is smaller which may result in frequencies in shifted to an area where the ear is very critical. The midrange enclosure may need to be tapered (which can be handily combined with diagonal bracing) or lengthened as in transmission line designs. This part of the enclosure can be filled with foam, or maybe even layered with different grades of foam.

Diffraction effects

The diffraction effects of the cabinet really only come into play once the distance from the unit to the edge of the cabinet is equal to a half wavelength. (Wavelength in centimetres = 34400 / frequency in hertz). As can be seen from this, unless the cabinet size is abnormally large, these effects are relevant mostly to HF units, and also to midrange units at the top end of their range. Diffraction effects will appear wherever the wavefront encounters sharp edges at a distance where the comparable wavelength is within the working range of the unit. The outcome will be a dip at the fundamental frequency followed by peaks and subsequent dips at increasing octaves (an octave being double the frequency). This explains why it is essential that HF units are flush mounted to avoid effects at the edge of the faceplate, especially as most HF units have round faceplates with the diaphragm central within it which concentrates the effects dramatically. Cone midrange units tend to suffer less due to the nature of cone behaviour, namely the tendency for piston movement to shrink radially with frequency, hence the outer part of the cone acts as a horn and increases directivity, meaning the the unit is very directional towards the top end of it's range and tends to not 'see' the edges of the cabinet. However, midrange units are usually placed close to HF units and are therefore usually flush mounted for this reason.

So, what can be done. In my experience, the closer the diffractive edge, the more severe the problem, hence small or narrow cabinets tend to suffer more. The severity of the problem is also increased with the angle of the diffractive edge. Conventional 90 degree edges show quite severe effects, but introducing a rounded edge or a good width 45 degree champfer to the front edges can show significant improvements. It seems that the units see these surfaces as bent flat planes. Taken to extremes, a spherical enclosure will exhibit almost perfect characteristics, but will suffer appalling internal standing wave effects due to all internal dimensions being equal. Spherical enclosures have been used successfully for HF only enclosures, but are hard to integrate into a finished design (Cabasse). It seems obvious that HF units should be mounted so that the distance from the top and sides are all different, but this is rarely apparent in commercial products. This is due mainly to the extra cost of producing two different baffles, although this is not a restriction for the home builder. It seems logical tho apply the aforementioned 'golden ratio' of  1 : 1.6 : 2.3.

It appears also that commercial products are designed, and then a grille is fitted. Very rarely is the grille integrated into the design with an assumption that when people listen to their speakers, they will remove the grilles as they 'obviously' interfere with the sound. I think that this is neither practical nor desirable. Sure some enthusiasts may do this, others may when they first purchase their speakers, but I don't believe the majority of purchasers do any such thing. The only real concession made to grilles is the good quality grille cloth available these days, as opposed to the older, more intrusive materials such as tygan, as fitted to the BBC designed LS3/5a. Talking of which, a method used on this speaker, and used on some other products, to improve HF performance is to fit strips of absorbant felt in a square around the HF unit. This works well but note that this felt is of a layered type, and only works when edge of the layers are placed so as to absorb the lateral output. It does not work at all if placed the other way round.

I would conclude by re-iterating that I do not present this information as anything other than my opinions, although some parts are undoubtedly fact. I have tried to keep this as short and simple as possible. If you disagree or think I may be able to help you, please e-mail me.