Technical Information

Enclosure Design Principles

    Non-parallel walls

    RR Audio designs enclosures based upon the fact that parallel walls cause standing waves inside the enclosure which cause certain frequencies to predominate at wavelengths based upon the distance between the walls. These frequencies and their harmonics will cause reinforcement (or cancellation) which causes peaks (or dips) in the frequency response. Depending upon the bandwidth of these resonances, there will be energy storage and ringing at these frequencies to some extent (a great deal if the resonances are high-Q). By designing the enclosures with non-parallel walls, the wavelengths between opposite walls are different at any two adjacent points, which spreads the frequencies associated with these wavelengths and eliminates standing wave modes. The reduction in midrange distortion and frequency response anomalies is significant. In addition, the rear-wave energy from the LF driver is bounced on paths which do not immediately re-contact the diaphragm. The energy is bounced all around the interior of the enclosure, encountering damping material at each wall which absorbs much of the energy, so when it finally recontacts the diaphragm, the amount of energy is greatly reduced from the original level. This reduces the phase distortion caused by rear-wave energy creating vibrations in the diaphragm which have no relation to the current audio signal, and substantially improves transient detail performance, low-level detail retrieval, and clarity.

    Panel Construction

    Enclosure panels are kept small to raise the panel resonance frequencies and reduce the amplitude of panel vibrations, reducing "cabinet talk" to a minimum. This improves midrange performance to a great degree. Larger enclosures which require correspondingly larger panels are braced in a way which reduces the panel vibration amplitude and raises panel resonance frequencies in the same manner.

    Waveguide shape

    A major cause of image defocussing is diffraction of the soundwave as it leaves the enclosure. This effect is similar to the spraying effect caused when water flows over the edge of a waterfall. As the water goes over the edge, it carries some of the air between the sheet of water and the cliff face with it, reducing the air pressure between the water and the cliff face. As the air pressure outside the sheet of water is greater than that under the water, it forces the water into the cliff and causes it to bounce off, creating spray (and some pretty neat rainbows when the sun is out). The identical phenomenon occurs when the soundwave which is traveling across the baffle (front face) of the speaker cabinet encounters the typical 90º edges of most loudspeakers. The soundwave bounces off of the side wall and re-radiates into the room (kind of like there was another little speaker at that position putting out the same signal at a lower level and time-shifted). These additional radiations cause "ghost images" which radically defocus the main image. By reducing the angle of incidence between the baffle and side walls, this spraying effect (diffraction) is minimized and the image is sharply focused and much wider, taller, and deeper. In addition, minimum baffle area is used while still supporting lower midrange frequencies on the enclosure.

    Enclosure tuning

    Enclosure tuning is accomplished using low-Q bass reflex techniques. This is a method of porting the enclosure that allows for wider bandwidths at the two resonant frequencies associated with the reflex (FL and FH), and correspondingly less peaking at these frequencies (and less energy storage and ringing which is associated with those peaks). Typical reflex tuning techniques yield a Q higher than 1.0 (Q of 1.0 = a peak of 3dB at the resonant frequency, and quite a bit of energy storage and ringing). This is a major cause of the "boomy", flabby bass associated with many reflex designs. By tuning for a lower Q, peaking and energy storage is eliminated. (The 2802 is designed with a 3dB peak at 42Hz, a frequency almost an octave below the area of most enclosure tuning peaks, to intentionally increase the bottom octave performance). Some of the enclosures are designed with two ports tuned at separate frequencies, a technique which allows a greater degree of control over the box tuning frequency, and a corresponding increase in control over the Q at FL.

    Damping Material

    Damping materials used in most loudspeakers are typically Dacron fiberfill, uncompressed fiberglas, or foam, and have little effect on low frequency energy absorption. The absorption coefficient of the materials used in our loudspeakers is quite high at low frequencies, which contributes to the ability to control the low-frequency performance of our products. Changing damping effectiveness alters the Q of the tuned system, the energy level inside the enclosure (and how much of that energy returns to the rear of the LF driver's diaphragm), and the frequencies of resonance at FH, FL, and FB (the high frequency impedance peak, low frequency I.P., and tuned frequency of the box). In addition, the use of certain materials actually reduces the amount of energy transmitted from the internal energy to the wall panels.

Filter Design Principles

    Parts selection

    Before discussing the basic principles we use in selecting a filter type - and to keep this readable to the non-engineer this will be in English :^) - a short discussion regarding filter parts themselves is in order. Filter (crossover) networks (high level) consist of inductors (coils), capacitors, and resistors and are designed to limit the operational bandwidth of a specific driver, set the parameters for circuit-Q at the tuned frequency(s) of the filter, arrange for smooth blending of the drivers, adjust for phase anomalies, etc. A properly designed filter will discharge any energy stored very quickly, pass as much of the signal through without absorption as possible, and allow for minimum overlap between the drivers while maintaining a smooth transition region to make the total spectrum as seamless as possible.

    There are many types of capacitors available on the market, made of different types of materials. The primary differences are dielectric absorption (how much of the signal gets through), dissipation (a function of the reciprocal of the DC resistance), and stability. We use metallized Polypropylene capacitors - not the most expensive material or the best dielectric (vacuum capacitors are the best, but are generally used only for very high voltages - and Teflon is a rather expensive material, although it IS better) but definitely in our opinion the best material at a reasonable price. In addition, some manufacturers capacitors are better than others (like anything else), so we test various types of caps and choose the ones we feel are the best we can find for the application.

    Inductors also come in different flavors :^) - such as Perfect-lay wire-wound (the cross-section and height are calculated to have the optimum core size and highest Q (effectiveness) for a given value with the lowest DC resistance) - and CFAC copper foil inductors which reduce the skin effect at higher frequencies (esp. 2nd and higher harmonics) and the resistive power loss. There are of course many options for inexpensive air-core coils, but we have found that it is worth the cost to use the better parts. We avoid iron-core coils, even though the resistance is lower, due to various losses in this type of part.

    Filter design

    There are many, many different families of filters, defined by the circuit topology and Q. We have tested and experimented with nearly 100 different types, and have chosen the filter families we use for their nearly perfect transient properties and phase characteristics. We avoid the 2nd order, used by most manufacturers, due to the 180° phase shift at the crossover point, and use 3rd order almost exclusively. The filter families we use are chosen for optimum selectivity (minimal overlap due to high secondary slope characteristics); linear phase performance; and little or no ripple in the frequency and phase response. We use two types primarily - Transitional Gaussian to 6dB and Linear Phase with Equiripple error (±0.05°) depending upon the requirements. These filter types give us the maximum amount of control over the drivers without causing any problems (such as ringing or overshoot, phase abnormalities, or frequency response anomalies). We typically adjust the phase characteristics of each filter to match the driver's requirements and to set up conditions of minimum overlap so that more massive, slower drivers are not operating over the same frequency range as lighter, quicker drivers. We also correct for inductive reactance in LF and MF drivers, so that the filter sees a stable resistance and operates as designed.

ML5a Redesign -- Filter network

The two-section filter on the right is a single channel for a studio monitor project which required that a pair of Manley d’Appolito monitors be gutted to just the enclosures, drivers replaced with top-grade material and a filter designed to spec. It shows the quality of parts used in a prototyping configuration, as well as the braided silver/teflon cable used for the internal connections. The filter on the left is the original Manley ML5a filter network.

Driver Selection Principles

    Low Frequency Drivers

    The LF drivers (woofers) we use are chosen for high stiffness-to-mass ratio in the diaphragm; high acceleration magnetic drive systems; rigid, non-resonant frame structures; linear frequency and phase response characteristics; and quality construction. A high stiffness-to-mass ratio is important because you want the driver's diaphragm to have as little inertia as possible (lowest mass) so it can react quickly, but you don't want that light diaphragm to flex. If the diaphragm flexes, the "bent" area becomes a little speaker which emits a different frequency than that being emitted by the rest of the diaphragm, causing distortion (and we want to avoid that, don't we?). A high-acceleration motor system will allow the driver to start and stop as rapidly as possible, delivering impact and detail which would be obscured otherwise. Use of a rigid, non-resonant frame structure ensures that a minimum of energy is transmitted from the frame to the enclosure walls, and that energy storage and ringing in the frame itself is kept down (the stiff frame has a low amplitude of vibration and a higher internal resonant frequency, like enclosure wall panels).

    Mid Frequency Drivers

    Midrange drivers are chosen for the same parameters as LF drivers, with the additional requirement that they have linear, extended HF performance with no phase anomalies within 1½ octaves of the crossover frequency to allow the smoothest possible transition with the HF driver. In addition, drivers are chosen with a resonant frequency at least 1½ octaves below the LF crossover point so that the resonance does not interfere with low frequency performance of the driver or interfere with the filter/driver stability.

    High Frequency Drivers

    High frequency drivers (tweeters) are again chosen for the same parameters as LF and MF drivers, with the additional requirement that they have response characteristics which are stable well beyond the20kHz limiting frequency of most human ears. We typically choose drivers which are extremely fast (to allow for the retrieval of the most low-level details) and very efficient, so that they will be able to handle the output levels required for realistic transmission of music without requiring an undue amount of amplifier power. We typically choose HF units that are a bit more efficient than the LF and MF drivers, and attenuate them to match, allowing the drivers to handle even more power.

    The HF drivers we are using are of two types, dynamic (dome type) and ribbon. The dynamic drivers are Titanium dioxide-coated Titanium diaphragms mounted on a non-resonant faceplate with a phase-correcting cone suspended in front of the diaphragm, and is an exceptional driver which uses a magnetic drive system larger and more powerful than many woofers. The Raven ribbon drivers use a magnetic drive system so powerful that you must keep any magnetically active materials far away from them, since it is very difficult to break the magnetic connection if they come in contact with each other! The diaphragms themselves are EXTREMELY light corrugated foil strips suspended in the magnetic gap, and are so fast that they must be heard to be believed. Diaphragms are easily replaceable in the field, by the way, unlike many other ribbon drivers, in case you take an unprotected speaker out into a high wind or something.

Analysis and Testing

    Low Frequency Drivers

    The LF drivers (woofers) we use are chosen for high stiffness-to-mass ratio in the diaphragm; high acceleration magnetic drive systems; rigid, non-resonant frame structures; linear frequency and phase response characteristics; and quality construction. A high stiffness-to-mass ratio is important because you want the driver's diaphragm to have as little inertia as possible (lowest mass) so it can react quickly, but you don't want that light diaphragm to flex. If the diaphragm flexes, the "bent" area becomes a little speaker which emits a different frequency than that being emitted by the rest of the diaphragm, causing distortion (and we want to avoid that, don't we?). A high-acceleration motor system will allow the driver to start and stop as rapidly as possible, delivering impact and detail which would be obscured otherwise. Use of a rigid, non-resonant frame structure ensures that a minimum of energy is transmitted from the frame to the enclosure walls, and that energy storage and ringing in the frame itself is kept down (the stiff frame has a low amplitude of vibration and a higher internal resonant frequency, like enclosure wall panels).

    Mid Frequency Drivers

    Midrange drivers are chosen for the same parameters as LF drivers, with the additional requirement that they have linear, extended HF performance with no phase anomalies within 1½ octaves of the crossover frequency to allow the smoothest possible transition with the HF driver. In addition, drivers are chosen with a resonant frequency at least 1½ octaves below the LF crossover point so that the resonance does not interfere with low frequency performance of the driver or interfere with the filter/driver stability.

    High Frequency Drivers

    High frequency drivers (tweeters) are again chosen for the same parameters as LF and MF drivers, with the additional requirement that they have response characteristics which are stable well beyond the20kHz limiting frequency of most human ears. We typically choose drivers which are extremely fast (to allow for the retrieval of the most low-level details) and very efficient, so that they will be able to handle the output levels required for realistic transmission of music without requiring an undue amount of amplifier power. We typically choose HF units that are a bit more efficient than the LF and MF drivers, and attenuate them to match, allowing the drivers to handle even more power.

    The HF drivers we are using are of two types, dynamic (dome type) and ribbon. The dynamic drivers are Titanium dioxide-coated Titanium diaphragms mounted on a non-resonant faceplate with a phase-correcting cone suspended in front of the diaphragm, and is an exceptional driver which uses a magnetic drive system larger and more powerful than many woofers. The Raven ribbon drivers use a magnetic drive system so powerful that you must keep any magnetically active materials far away from them, since it is very difficult to break the magnetic connection if they come in contact with each other! The diaphragms themselves are EXTREMELY light corrugated foil strips suspended in the magnetic gap, and are so fast that they must be heard to be believed. Diaphragms are easily replaceable in the field, by the way, unlike many other ribbon drivers, in case you take an unprotected speaker out into a high wind or something.



If you would like to ask questions regarding the products and services we offer, or would like us to send you more information, please let us know. We would be happy to provide you with whatever information you require.

RR Audio Laboratory
636 E. Harvard Rd. unit B
Burbank, CA  91501  USA
Telephone : (818) 843-8212
Facsimile : (818) 563-9372
E-mail : 
rr@trapagon.com