General thoughts about Loudspeakers

Introduction
The first, and maybe most important words in this (long!) article, are the following: This is my personal opinion! I do not claim to tell absolute truths, we are all different as humans, we have our own personal reasons for listening to music and for selecting the equipment we use. The joy of listening to music of any style or genre is the fundamental, paramount core, and all the technique we use as help are only stepstones on this path.
My own reason for building loudspeakers the way I do is clearly related to my work as a professional musician, my work with mixing and mastering my own music, and 16 years as writer, equipment tester a.o. at the Scandinavian magazine High Fidelity In other words: I want to listen to the music material itself, in all it's glory, and with it's flaws as distortion, bad recordings, bad circuitry and microphones etc. I don' want to hear my equipment, or disturbing interference from my listening room. For me, this is the way I have chosen to walk.



TLL

JUNIOR

BABY

Special Woofer Design

Proof of Concept

The following text is the more general description, I will add a file with more theory, data sheets and diagrams later on.
Loudspeaker Cabinets.
Let's start with cabinet type, I will focus on closed cabinets and bass reflex cabinets, since these are the 2 types I have been working with. For maximum bass reproduction accuracy, there is only one option; the closed cabinet. With a bass reflex cabinet, you create a resonance system with the internal cabinet volume and the bass reflex port. This resonance affects the impulse response in a negative way, and yes, the difference is clear. So, with a closed cabinet tuned to obtain the best possible impulse response, you get the most precise reproduction of bass frequencies, and low freq. Impulse sources, like a kick drum.
Drawbacks: limited choice of woofers, since the majority of the commercial design are bass reflex types, and manufacturers of woofers/loudspeaker units as a consequence have to focus on this type. The difference lies in the physical and electrical parameters of the speaker. Solid and big cabinets, the internal air pressure inside a closed cabinet is bigger, compared to a Bass Reflex. For excellent impulse behavior, you need a low Q factor, and this means a big box! One example is my TLL loudspeaker, even with my truly excellent 2 Sinus drivers (Isobaric mounted, which gives a lower Q factor in itself) I need 140 liters of volume to achieve a Q of approx. 0.56. (More about Q factors in the technical section).
A well designed bass reflex cabinet is option nr. 2. If you design the tuning – cabinet size and port dimensions – with care, it's an acceptable compromise between frequency response and impulse response. Avoid tunings with maximum bass output as target, the result will be something referred to (in the old days) as one-note bass or boom-boom bass, especially if your listening room is on the resonant side.
Cabinet shape
Avoid parallel surfaces! My Baby design is an example. Naturally, this means a more complex cabinet design, but there is an additional benefit, you get a more robust structure, compared to the classical box shape. The reason for avoiding parallel surfaces is standing waves, that builds up inside the cabinet, and amplify certain frequencies related to the distance between cabinet walls. With a non-parallel design, the effect is reduced.
Keep the front slim to minimize baffle effects. Round edges to minimize diffraction. Try to place the speaker units close to each other, to get a better illusion of a point-source sound radiator, also to reduce phase-shift influence on the frequency response due to the difference in distance between speaker units and the listener – this problem also relates to the crossover design – see that section!
Cabinet materials
General: go strong and heavy! Add extra material/bracing in cabinet corners, especially closed cabinets have to be air-tight! Double-wall cabinets with sand in between is luxury, but also a heavy and complicated solution. Avoid thin, resonant materials. Experiment with a sandwich of 2 different materials! Combine materials with different density. An easy way to check is to cut 2 identical squares of both materials, hang them up in a thin thread, hit them with a small hammer, record the sound with a simple microphone (keep the same distance to the square!)connected to a computer, and record the sound, (or record it using your mobile phone). Glue the 2 squares together, and perform the test again. Afterwards, check the result using a freeware software program such as Audacity or similar. Normally the result is obvious, the vibration in the sandwich dies much faster, and also faster than a square of either material in double thickness. The front board of the cabinet (housing the speaker units) can be twice (or even more) in thickness, compared to the other cabinet walls.
How many loudspeaker units?
2-way, 3-way, 2-way with subwoofer, 4-way.... - where to go? For the best overall sound reproduction, I recommend a 3-way system with a woofer designed to handle really deep and powerful bass. I also prefer a closed box system – for several reasons – more later on. In TLL, I had access to truly excellent 12” woofers, they do an outstanding job regarding all aspects. Deep bass reproduction, power handling, transient response, linearity and distortion. Drawback is price, 2 top-of-the-line (I use 4 in TLL...) 12” woofers are expensive. 10” is a fine alternative, 8” is for my taste and choice of music on the small side, power handling and real low bass reproduction begin to be reduced. 6.5” is for me too small – if you want them, use 2, and choose bass reflex cabinets as I have done in Junior.
A 2-way speaker will always be a compromise. Consider your listening preferences, if we talk softer, acoustic music, classical chamber music or similar, it can be a fine choice, mainly for smaller rooms. For classical symphonic music at realistic levels, or Heavy Metal at +100 dB – it is not going to work.
2-way with subwoofer? Has never been my cup of tea, they are seriously hard to integrate in the soundstage. Their impulse response is most likely to be different from the main speakers, and often they go higher in frequency than intended. This spells phase problems affecting frequency response and stereo image problems. They also have a tendency to activate room resonances. In the end, the solution will also cost some money. My advice is to use better 2-way speaker system, or to step up to a real 3-way system.
4-way?? Maybe for a really large room, where you need lots of power, and listen at some distance from the speaker. In more normal rooms, I find that the point source illusion is lost, and due to the variation in listening distance from each speaker unit, the frequency response suffers a little. It can also end up being quite expensive.
Conclusion on my side: 3-way systems are my preferred choice, good 2-way systems can be an acceptable compromise if it's well designed.
Crossovers
Many pages has been written about crossovers, type, (Bessel, Butterworth, L-R etc.) order, (1:st, second, 4:th) passive, active, analog, digital – a long list. I have worked with more or less everything. Passive: most types from 1.st to 4:th order, active analog as normal IC based and my own gyrator version. Digital as IIR and FIR. This is my platform for the following general discussion.
In the beginning there was the passive filter. I have mainly used second and fourth order as L-R type. Fourth order is fine for protecting your speaker units, but impulse response is far from perfect. Large filter coils for the woofer means added resistance, and loss of woofer control as seen from the amplifier. With top level components it can get expensive. I ended with L-R second order, as a fair compromise between unit protection and impulse response. Fewer filter components, less coil resistance.
Generally; measure all components, tolerances should be 1 percent or less, in order to secure equal conditions for all speaker units in a stereo system. Try to keep coil resistance low. Capacitors: I have a personal weakness for the old paper-in-oil type. They are typically big, and the construction secures very low thermal capacitance variation. Voltage ratings are often high, up to 400 Volts or similar, no fried caps when you turn up the volume! Measure them, and parallel with smaller types to reach correct value.
For a period, I used (on TLL) a combination, active analog filter for the woofer and low midrange, passive for high midrange and tweeter frequencies. Everything has to be checked with an acoustic measurement system, and if you do, there is no problem, worked fine. In a sensible listening room there is no question, the absence of woofer crossover filter coil in the signal path can be heard. More definition, more control. With controlled components in the passive filter, crossover frequencies were identical in the passive and active filter.
For a number of years, I have worked with digital filters, both IIR and FIR. This gives great freedom for a loudspeaker constructor, many filter types and orders (with FIR, there are no real orders), and delay and output power options. Compared to passive filters, with damping pads and restrictions when it comes to choice of loudspeaker unit, it is so much easier. Woofer control is superb, find a power amplifier with muscles, and you can handle big and difficult woofers with ease. Today, I use a power amplifier for every loudspeaker unit in my systems. Yes, it costs, but sound improvement is clear. Power requirements are very different when you compare for instance woofer and tweeter, the tweeter is normally an easy load, and requires much less power. Sound reproduction is another issue. Often, large power amplifiers (+200 W or more) are not the best choice for smooth tweeter sound reproduction, a smaller and more delicate amp can provide a more beautiful and realistic sound.
Choice of crossover frequencies. An equal amount of pages has been written about this subject! General guidelines are easy: check manufacturers data sheet, you will often find recommendations, to protect loudspeaker units from possible damage. From my perspective, these limits makes good sense.
My own choice of crossover frequency depends on a number of factors: safety limits, size of unit, frequency response including on – and off axis response, cone excursion combined with power handling. Loudspeaker size matters!! I don't want a high crossover frequency (cf) for a large woofer, cone breakup, reduced dispersion, more distortion are typical drawbacks. Neither a low cf for a smaller midrange – large cone excursion, more distortion and stress for the unit. The same applies for the high end – don't push the mid too high, don't set the tweeter cf too low. When we listen, we hear not only the on-axis response, we hear the total response of the loudspeaker unit. This means that a sudden change in dispersion will be noticed, as a disturbance or something unnatural. An example could be a 2-way loudspeaker where the woofer is pushed very high, and hence more directive, and a tweeter with wide dispersion at the same cf.
Damping Material
Damping material is not as easy as it seems. All damping materials have different acoustic properties. From an ideal point of view, it should work as a black hole, the air should just disappear into it completely. Technically/theoretically speaking, air movement energy should be transferred to heat. But how to make this happen in the real world? The amount of damping material inside a closed cabinet matters. In a front board, as can be seen on my TLL and Junior cabinets, thickness matters. General rule says that damping material starts to get effective at about ¼ wavelength. Example: front board/pad, thickness 5 cm. 4 x wavelength is 20 cm, equal to about 342/0,2 (speed of air / wavelength) – 1710 Hz. This gives an indication of what you can expect, a board/pad is good for tweeters, gives some, but less, effect for midrange and almost useless for woofers. You can see the impressive effect on my measurements on TLL tweeter Link to TLL Details Midrange foam pad has much less effect, also when measured.
This brings me back to the starting point, are damping materials perfect? So far, I can say no. There are 2 important parameters, frequency linearity and amplitude linearity. The first is easy to understand, all frequencies should be absorbed by the damping material, this is what we hope takes place inside our well-filled closed box. Does wavelength matter – is sound absorbed just as efficient at 30 Hz as 450?? I hope it is, but I can't say I am sure, since it is hard to measure. Which brings the amplitude parameter in play, does it matter, if sound pressure is 68 or 89 dB:s?? Is the damping material linear, or is there a point of overfill, where the material can't absorb more, and starts to reflect the sound wave instead of absorbing it?? My present listening tests with foam fronts indicate that overfill is possible, and as a consequence, I am looking for alternatives. Does anyone know if there is any serious research in this area??
For woofer closed cabinets, I have used rockwool and glasswool, can't hear or measure any difference. For midrange cabinets, I prefer wool, long haired sheep wool or similar, for some reason it feels/sounds better than rockwool, can't say why. Anyway, be very careful to use exactly the same amount in left and right speaker cabinets – this is important, for resulting resonance frequency and for Q factor / impulse response behavior. With my experience, midrange Q factor is very important; too high, and precision gets lost, this is an area where we hear really well. My personal preference is more or less as low as possible, 0.5 is super, 0.6 ok, 0.7 top limit. And as said before, midrange cabinets should be solid!!