This review page is supported in part by the sponsors whose ad banners are displayed below

Antti Louhivaara: My current work is mostly a natural progression of the time I spent grappling with these issues at Amphion. There I concentrated on controlling radiation patterns in several ways. Waveguides were used to adapt tweeter radiation patterns to midbass units as well as maintain smooth phase response at the crossover frequency. Of course the main goal was always to increase the directivity of the total acoustic system. Therefore I used resistance boxes to extend directivity into lower frequencies, this mostly with Amphion's Xenon. With the Krypton this theme was developed even further by increasing vertical directivity with D'Appolito-configured midranges in special resistance sub enclosures.

At Aurelia we now do things in mostly similar fashion, albeit with certain advances even in our basic Magenta, Ambera and the upcoming Saphira. My current ECW waveguide is pretty much similar to what Amphion used in most respects. We did improve the throat shaping for a smoother energy transfer. Our woofers however are quite a bit different than what we used during my Amphion days. I am very proud of this new DDC woofer. It really works. Major differences to what I've done before come into play with our Cerica and Graphica models. I learned just a few years ago that increasing vertical directivity improved the sound over anything I'd focused on previously. I therefore decided to continue with cylinder wave radiators. I began to develop different acoustical systems based on that principle. One other issue I have now realized is the importance of time coherence. Music happens in the time domain of course. This should be kept in mind when designing speakers. Now I'll explain my current design approach in more detail.

In the 1970s, Joseph D'Appolito created an acoustical array of two mid/woofers symmetrically arranged above and below the tweeter in a vertical alignment. The main purpose was to eliminate lobing errors in multi-way speakers. Increasing vertical directivity wasn't initially important. Under certain conditions where poor listening room acoustics normally ruined the sound, one soon observed improved imaging and sound quality with this alignment. After more critical investigations, certain problems became evident as well. A D'Appolito system sounded vertically unstable and vague with even the slightest movement in the listening position. Thought the D'Appolito array today still has its supporters, most manufacturers ignore its strengths due to its compromises. The main problems with the D'Appolito configuration are:

• Interference at the crossover frequency: In most cases the distance between the radiating surfaces of the mid/woofers is larger than the crossover frequency's wavelength. This causes strong interference while moving in the vertical axis. Once you add a tweeter into this mess, the result is an acoustically very unstable behavior in the vertical axis.

• Different kinds of waveform behavior between tweeter and mid/woofers: Personally I find the even bigger problem to be that in a traditional D'Appolito configuration, the tweeter operates as a spherical radiator whilst the two symmetrical midranges tend to generate cylinder waves at the crossover frequency. While a cylinder wave fades at 3dB over distance, this doubles with a spherical wave to 6dB. The result is output instability over listening distance. To correctly balance a D'Appolito system is pretty difficult. Even in the best case the result will be more or less unstable. Is there a way to get a D'Appolito array to work properly?

The only way to avoid interference at the crossover frequency is lowering the crossover point so that the distance between the radiating surfaces of the mid/woofers is shorter than the crossover wavelength. In most cases this means frequencies below 2kHz. The mutual lack of coherence between tweeter and midranges is solved with radiators which generate cylinder waveforms proportional to the size of the diaphragms. Tweeter and midrange systems both should generate cylindrical waves at the crossover frequency.

When these two requirements are applied simultaneously, we avoid instability. This is no easy task in the real world. Though midrange drivers are moderately easy to get working correctly and generate coherent cylinder waves at the crossover point, the problems arise with the tweeter system. Traditional ribbons certainly generate cylinder waves and there exist a fair number of ribbons with the right length to properly integrate. The problem with ribbon tweeters is their poor power handling at lower frequencies. This enforces a crossover point that is too high. In addition, the steep slopes mandatory to protect ribbons create their own problems. For these reasons, I began searching for a solution which would differ significantly from existing ones.

Many years ago I explored different ribbon tweeters from Aurum Cantus, Fountek etc. in my prototypes. Without steep high-pass filtering, they all sounded very good. The sound was transparent, the 3-D effect excellent. The problem was the almost non-existent power handling and/or the crossover frequency being too high. Whenever I inserted the mandatory steep high-pass filter, something sonically essential disappeared.

This had me think about a solution which would generate waves like ribbons but with traditional dome tweeters. It would improve power handling to avoid a steep high pass. If the radiating surfaces were close enough, interference problems could be avoided while simultaneously generating similar cylinder waves to ribbon tweeters. After many experiments and prototypes I ended up with the current solution. It combines three neodymium domes in a shallow waveguide. Its throat is carefully designed to avoid mutual resonances and integrate the domes to work together as uniformly as possible. The radiating surfaces are close enough together to render interference artifacts insignificant at audible frequencies. Distortion levels are lower than with virtually any ribbon tweeter currently available. The sound remains very "ribbonish" however and because of the lack of a steep high-pass filter, it maintains all the spectacular 3-D information and spaciousness. Sensitivity at 97dB/Wm is pretty high.

After my experiments with ribbons and my CSR cylinder source radiator prototypes, it became evident that the so-called ribbon sound stems primarily from how it generates cylinder waves, not low mass or anything else on the ribbon technology topic. My CSR triple-tweeter system forms cylinder waves above 1.5kHz. That also is the system crossover frequency. As I said earlier, this lower crossover point is necessary to avoid interference problems at the crossover frequency. When talking about line sources, it is necessary that both mid/woofer and tweeter system generate uniform cylinder waves to obtain stable and uniform acoustical behavior over the entire acoustic system. Both my Aurelia Graphica and Cerica models realize these ideals.

When talking about real cylinder radiators (not WMTMW solutions), a single dome tweeter won't work satisfactorily as a line source. There is no sense in combining cylindrical and spherical radiators. Plus, it is very difficult to integrate a dome so that its radiating impedance remains as resistive as possible. To do that requires a pretty high crossover frequency which immediately causes interference issues with the mid/woofers. I thus emphasize that the only reasonable way to get satisfactory results is to use cylindrical radiators for all drive systems which must be used at the frequencies where their radiating impedances remain as resistive as possible, i.e. radiator lengths are proportional to the frequencies they generate. This is the only way to get a uniform radiation pattern and a smooth energy response and stability as a function of listener distance.

When these requirements are dealt with, the sound becomes stable both vertically and as a function of distance. The horizontal dispersion remains pretty similar to traditional solutions. Because the drivers generate cylinder waves over only limited frequency ranges, there exist other kinds of problems whose annoyance is more or less a function of taste. My CSR tweeter system is vertically very directional, hence the upper frequencies tend to vanish above the typical listening height. However, this happens without the kind of vagueness associated with traditional D'Appolito configurations. Outside the nearfield (>4m in the case of the Graphica), the upper frequencies integrate seamlessly.

Again, this type of speaker is vertically very directional. Like panel speakers, the Graphica has one prime spot for listening where it offers excellent imaging. Outside that spot the sound changes considerably. Therefore Graphica and even Cerica never pretend at being any kind of universal solution. That said, their smooth energy response guarantees reasonable results even when listening to music in the kitchen whilst making sandwiches.

Both Graphica and Cerica use 5.25" mid/woofers. This limits the kind of LF dynamics possible with larger woofers even when using paralleled drivers (six in the Graphica, two in the Cerica). Our goal thus became to maintain the best possible dynamics and slam in the band where these drivers work optimally. In the real world, Graphica's bass response extends to approximately 30Hz, Cerica's to 40Hz in a typical listening room. Because of the dimensions of these sound-radiating systems, best driver integration occurs at a minimum listening distance of 2 meters in the case of Graphica and 1.5m in the case of Cerica. With Graphica you are listening in the nearfield up to 4 meters.

Because Cerica's radiating array is significantly shorter (Cerica doesn't operate as a cylindrical radiator below 400Hz), its nearfield doesn't extend as far as her bigger sister. If you can live with these compromises, both Graphica and Cerica offer exceptionally high sound quality in typical listening rooms.

While designing our current tweeters and mid/woofers, I strongly emphasized their natural compatibility with each other by matching their off-axis responses. The Graphica thus actually needs no crossover at all. Without a crossover the midrange would be emphasized over a wide range but would work pretty well. All we finally needed was a simple 1st-order filter on both tweeter and midbass arrays. It is one of the reasons why time coherence of the Graphica is one of the best in the market. Speed and responsiveness are actually on the same level as single-driver widebanders.

Today it is very common for typical mid/woofers to have linear excursion of 12-16mm even with small diameters. Bigger drivers can exceed those values by a factor of three or more. Air gap length is usually 6-8mm which leaves 50% or more of the voice coil outside the gap's magnetic field. If more than 50% of the voice coil sits idle, it typically creates 3 - 4 ohms outside the gap to present significant series resistance to the speaker cable. This has a drastic effect on the damping factor of the system and compromises the driver's ability to control cone movement. In our DDC units, linear excursion is purposely smaller than usual to decrease resistance loss to below 1 ohm. Your amplifier's ability to maintain control over the cone movement is hugely improved over the typical woofers used nowadays. Connected as an 8-ohm device, the BL of our mid/woofers is 12. That's a huge amount for transducers their size.

Introductory conclusion: The Cerica operates as a line source down to 400Hz. Below that it becomes a spherical/omni radiator. The cylindrical wave propagation above 400Hz creates an optimized listening window which is very focused in the height domain. That imposes significant treble attenuation when standing up. The sound field coheres best at 1.5m or higher distances to make most desk-top listening slightly less than ideal. The single-port paralleled woofers run a variation on the underhung voice coil scheme to increase amplifier damping. The crossover slopes are a 1st-order 6dB/oct. high-pass on the tweeters, a 1½-order 9dB/octave low-pass on the mid/woofers.