Full-range Driver & Loudspeaker Theory
Full-range (or wide-range) drivers come in two general types: the center cap and the whizzer cone. The center cap type looks like your normal cone speaker with a dome center cap. The whizzer cone type has two cones, a large cone (as in normal cone speakers) and a smaller whizzer cone inside the large cone. Driver units have quite different characteristics depending on the design and manufacturer. Some (like Lowther) have very high efficiency, low excursion, and a frequency response graph that shows lowered sound pressure levels at medium to low bass (below 400 Hz). For this reason, to be an effective full range speaker, these drivers usually need special cabinet construction designed to enhance the bass output, such as a rear-loaded horn. However, the high efficiency allows the use of very low power amplifiers (such as a two Watt 2A3 tube amp). Other drivers have lower efficiency but flatter frequency response. These drivers can often be used in an open baffle or bass reflex cabinet, but require higher amplifier power to drive them.
There are very few true full range drivers that radiate from one source. The Diatones almost qualify, but are a bit weak in the bass, and a little weak in the highs. Some old drivers such as the Goodmans Axiom 80s, and maybe Telefunkens and Norelcos and others go down to low bass (30s) and pretty high (12 - 16 kHz) making them more full range. These all radiate "from the front cone", making a true point source and having no or little time delay errors. Finding or making these drivers is very difficult. Most "full-range" drivers end up needing two radiation sources, one for low and one for high frequencies.
Lowther drivers, for example, radiate the high and mid frequencies from the cone, but the bass come out of a horn and are delayed in time and 180 degrees out of phase, as well as coming from the horn mouth rather than from the cone. So in a way, you have two frequency sources, with a crossover between them. The crossover is acoustic and needs to be tuned properly, and at the crossover frequency there is still potential for interaction of the two wave fronts.
Another option with a Lowther-style radiator (which goes from high to mid frequencies) is to add a separate bass system. Bert Doppenberg's Oris 150 front horn for Lowthers needs a separate bass system, Bert uses bass horns with dedicated drivers. Subwoofers might work but are often not "fast" enough to integrate properly. This philosophy puts the (one) crossover at the low frequencies (150 - 200 Hz for Bert's, approx 300 Hz for most rear-loaded Lowther drivers).
Another philosophy puts the crossover frequency high up, at 12 kHz, or 15 kHz, the higher the better. This method uses a driver that runs from low bass (30 or 40 Hz say) all the way up to the crossover frequency (10.5 kHz say), then lets a supertweeter take over from there. The big question is, where is it better to put a crossover? Low frequencies or high?
A second question is, what kind of bass do you like? Horn-loaded bass (whether from a Lowther horn or a dedicated bass horn) is regarded as being "fast", having good transient response characteristics, a very natural sound. Direct radiators have a harder time making that kind of bass. Bass drivers with long throws and polypropylene cones are particularly bad. A bass driver cone made from paper (or sometimes kevlar or other modern material) often sounds better. But a full-range driver such as a Diatone cannot have too large a diaphragm and also cannot have too large a cone movement (xMax) because otherwise high frequencies will be attenuated and intermodulation distortion will result from the long throw. However, to make more bass, you can increase the surface area (cone size) without increasing the travel (xMax). Now you can have more bass and higher treble without too much intermodulation distortion, as long as you can control the larger cone (need good design for the cone). With the shorter travel, bass will become more controlled, crisper and more responsive compared to a long travel bass driver. (You won't get as much bass out, but it will sound nice).
Cone material can be made from paper (the classic material), plastic (often polypropylene), synthetic fiber/resin such as kevlar, and metal (often aluminium). Paper tends to have less stiffness than some other materials (kevlar and metal) but doesn't exhibit the ringing resonance often found in kevlar and metal. Some paper cones can be stiffened up with varnish-like substances such as Damar. Good discussion on driver theory is by Bob Stout at LDSG. Some excellent articles by Lynn Olson on The Art of Speaker Design and speaker drivers (along with some tube thoughts).
Some thoughts and theories on imaging of speakers are here by Jeff Bagby. Single driver speakers appear to be excellent, with Jordan JX92s as an example.
Physics of Full Range Drivers
By John Wyckoff
Any attempt to explain full range cone-and-coil loudspeakers must begin with the fact that there is no such thing. Rather, such drivers are systems of two or more drivers on the same motor assembly. This may seem to be splitting hairs, but it is a very important concept in understanding drivers designed to operate full range. The two basic formats for full range drivers are whizzer cone and center cap type. In either case, there is going to be a substantial peak in the frequency response where the main cone and secondary device are both active. This spike is most easily understood in the case of the central dome type driver, such as the JBL LE8T-H. If a measurement is taken of the physical width of the dome, and that measurement is then converted to a wavelength, it may be seen that the co-resonance of the main cone and the dome is 6.1k Hz.. A measurement microphone will confirm this. This spike is fairly narrow in Q, but about +7 dB in height. A secondary resonance is created at the half wave point as well, or where the dome equals one half of 3.05 kHz. This resonance is lower in amplitude, and can be heard as "glare". In the case of the whizzer, or free cone driver, the exact point of resonance is more difficult to find with a ruler, but is still just as prevalent.
A free cone has many additional problems for the designer to overcome when compared to a center cap type. Free cones have radiating surfaces which are not in phase with each other, nor do the inner and outer surfaces receive the same air loading ( I've measured free cones which have 90 degree phase shift). The inside of the free cone is its own conical horn, while the outside is horn loaded by the main cone, which is an entirely different horn formulation. Anomalies also result from the outer edge of the free cone as the front and rear waves collide along this diffractive edge.
When the above difficulties are added to the other resonances within the driver; such as: fundamental resonance, cone resonance, and free-cone or dome resonance, the design job begins to look impossible. While good quality broadband drivers do exist, finding the right compromises is one of the most difficult problems in audio. For example, if enough mass is added to the main cone to reduce the cone resonance, the higher frequencies will be attenuated. If the cone is too light, it probably won't be strong enough to act as much of a piston at low frequencies. A driver which has enough excursion to offer dynamic bass, will exhibit higher intermodulation distortion. Full range driver systems are not a bass driver with a whizzer.
As mentioned before, there are more anomalies with the free cone approach than with a central dome, yet free cones predominate. This is because they offer much greater dispersion, and more adjustability. The Hammer Dynamics Super 12 drivers are of free cone and center dome design, to extend the bandwidth as much as possible. This third device offers challenges of its own.
Many smaller broadband drivers (200mm and under) are rear loaded into horns, in order to extend bass. There are several problems related to this type of loading: Size, delay, coloration, and the fact that this approach is the application of feedback. Striving to produce amplifiers without feedback, and then introducing it into the speaker design, seems odd engineering.
by MG (from a Full Range Forum post)
Whizzer cones were originally an afterthought to extend the high frequency response of midrange cones. Midrange paper cones absorb the flexural wave as it travels out from the voice coil. This means that there is a reduction in radiating area and high frequency response. Since the wave speed is proportional to the square root of frequency this usually ends up being a 1st order roll off, but please don't ask me to explain this in detail in a simple posting!
The midrange of paper cones can be extended by flaring the cone, so that the centre is mildly horn loaded, this gets you up to about 5kHz. The whizzer is flared to fit inside the main cone, and can sometimes be be used to horn load the area between the whizzer and the cone. Normal DIY practice is to fiber wool fill this region to avoid higher frequency cancellation.
By being light and stiff, the whizzer does not greatly affect the working of the main cone, the flexural wave traveling out from the voice coil pretty much unattenuated. This means that the high frequency can now be radiated over a much larger area, and hence the response is extended. The front of the whizzer does not horn load the centre, although the shape will affect high frequency dispersion.
Since the whizzer is unconstrained around the edges some cone modes can occur, although half roll surrounds do not damp these out as much as people would like to believe. The best trick for this is to silicon fluid damp both paper cones - since the whizzer is much thinner it will absorb much less silicon fluid, so its high frequency response will not be greatly attenuated. Critical damping of the spider is also very successful at eliminating these, but can increase the mechanical damping Qms parameter.
Understanding of diaphragm flexure has led to a new generation of whizzer-less full range designs, the pinnacle being the Jordan JX series. For my money these represent the ultimate cone flexure technology, but there are many other available drivers. Increasing practice is to use the same passive electrical network, used for baffle step correction, to extend the high frequency response - although this is limited to about 1/2 an octave.
Martin Garrish website featuring his home-built full range driver, with description and theory.