Having already written too much about amplifiers I decided, for a change, to add a speaker design section. None of the ideas tried are intended to be highly original, this is just a 'fun project'. An experimental approach will be used, with a minimum of theory and analysis. The idea is to try different options to see how they sound and measure before deciding what works best. (i.e. 'trial and error').
To keep cost and complexity low I wanted to use a full-range drive unit. A previous enclosure I tried had space for only an 8cm unit, so although that design was eventually abandoned it determined the speaker chosen. Fortunately there are a few good quality 8cm full-range drive units available from Visaton, Tangband, Fostex, Aurasound, etc. Some of these are too expensive to qualify for a 'low cost' project, but the Visaton FRS8 (4ohms, model 2003), while unlikely to be the best possible, is both cheap and highly regarded. Some of the others need notch filters to reduce response peaks but, if the manufacturer's specifications can be trusted, the Visaton is reasonably flat up to 10kHz without such additions, with a few moderate peaks and dips from 10kHz to 20kHz where the effect should be less audible. There may have been a change of specification, with the current Visaton website (Oct. 2006) listing resonant frequency 115Hz, Vas 1L, and Qts 0.69, while other sources, possibly out of date, say 130Hz, 0.8L, and 1.07 respectively.
Here is the Visaton FRS8, model no.2003, bought from BMM Electronics in Holland. There are various 'tweaks' which could be applied before fitting the speaker, for example applying a thick coating of 'Blu-Tak' or 'Sello-Tak' to the speaker frame to damp its resonances, as shown above on the right. An example of this and other possibilities were shown on DiyAudio. Using a screwdriver to tap the outer metal edge of the speaker with damped frame, it sounded very dead in comparison to an unmodified speaker. This is no guarantee that there will be any audible improvement in the finished design, but it seems unlikely to do any harm, and the Sellotak only cost 99p for enough to treat at least five speakers.
Measuring resonant frequency, fs, of this speaker gave a result near to 150Hz, higher than either the old 130Hz or the new 115Hz specification. It can be expected that the value will fall after extended use, but this was still a little discouraging. I decided to leave the speaker with a 2V rms 20Hz sine-wave applied for a few hours and retest to see the extent of the fall. I have no reason to believe 20Hz is the best frequency to use, but it can give a large cone movement while being inaudible. After 3 hours fs was 128Hz, after 4 hours 120Hz, and finally after 5 hours the specified figure of 115Hz. It occurred to me that the temperature of the suspension will rise with extended use, which may cause a change in fs, so I left the speaker disconnected for an hour and measured again, now finding fs to have risen to 121Hz. Whatever the reason, fs is clearly somewhat variable, and there is little point trying to be too exact about its value, though it may settle down to a more constant frequency eventually. This sort of variation is well known, not unique to the FRS8.
A rectangular sealed wooden box used as an enclosure has some well known problems, for example the flat parallel internal surfaces can cause resonances, and the sharp edges can cause diffraction effects. Using a curved enclosure can reduce these effects, but making one is difficult, so for the first project I started looking for something ready-made which could be adapted.
The first possibility I found is a small wooden fruit-bowl, bought from our local Morrison's supermarket for only £2.99 (UK). The maximum outside diameter is 8 inches (20.4 cm) and the height 5.5 inches (14 cm), the wall thickness is about 7/16 inch (1 cm) and the internal volume 2.2 litres.
The first idea was to cut a circular hole in the base to mount a suitable drive unit, but initial experiments resulted in such an unpleasant sound that this was quickly abandoned, and instead a flat front panel used, with a bowl forming the back of the enclosure.
Since then I also found hollow steel spheres available from Homebase, with diameters 24cm and 32cm, giving volumes 7.2 and 17 litres respectively. (The smaller size was £12.99.) These appear to be intended as garden ornaments, and have a small flat section for them to stand on, which could be cut out to fix a fullrange driver. A few examples of spherical metal enclosures can be found, e.g. on DiyAudio, but a wooden enclosure is more suitable for the sort of experiments I wanted to try, so the first project used the wooden bowl as part of a 'distributed port' enclosure :
A Distributed Port Speaker.