Speakers, rooms, and speaker sound.
First Topic. OK, my latest do-it-yourself project regarding speaker building is complete. Previously my main A/V system had a home-built center speaker that consisted of four Roy Allison designed AV-1 minispeakers assembled into a stand-mounted, triangular shaped configuration resembling the top half (top two feet) of either of my Allison IC-20 main-channel speakers. There were two front panels, each angled 45 degrees to the side from straight ahead, with a vertical MTTM driver array on each of them. The tweeters were Allison designs, but the 4.5-inch cone midrange drivers were built in the Far East for incorporation by Allison into the AV-1 units. The crossovers were basic second-order designs (high and low pass filtering) that Allison has traditionally used in many of his systems.
The four 4.5-inch cones notwithstanding, this system was unable to go deep enough into the bass range to be operated full bandwidth, even as a center-channel speaker. However, rather than configure the center circuitry of my receiver to route its bass to my main system's Velodyne F1800II "main" subwoofer, I gave the channel its own dedicated subwoofer: basically a modified (by me) SVS 16-42 unit, powered and crossover controlled by a Hsu Research amp. The resulting center-channel performance was pretty good, but the system had several weaknesses.
First, although the Allison tweeter is a great one, with a wide-angled radiation pattern in the top octave second to none, it is at its happiest if crossed over above 3.5 kHz. In the two-way Allison models, including the AV-1, the crossover point of the Ferrofluid cooled version of the tweeter is nearly an octave lower, at 2 kHz, and down that low the tweeter has the potential to have a bit more distortion with high-powered inputs than some might like. Although using four of the things in the array certainly helps out, I still prefer a higher crossover point, if only for power-handling advantages in the 2 kHz to 4 kHz range where a typically more solidly built midrange driver would be working in three-way Allison systems.
Second, the off-the-shelf midrange driver used in the AV-1itself is problematic. The unit is OK, but if I fed single-frequency test tones into it at moderately high levels I could hear harmonic artifacts that were not showing up if I fed identical signals into my IC-20 systems. Its inherent smoothness was not all that great, either, and the genuine, "dedicated crossover" controlled Allison midrange drivers in the IC-20 units were superior to the AV-1 midrange units both in terms of distortion and smoothness. Also, with a subwoofer crossover point of 90 Hz, I felt that the four AV-1 midrange drivers were at times having to do more high-output work in the middle-bass range from 90 to, say, 150 Hz than they were comfortable with.
The replacement system uses drivers and crossovers that I pulled from a pair of no longer needed three-way Allison AL-125 systems some time back. This system, a floor-standing design, has a single forward-facing panel, with a vertically oriented MTTM driver array, eliminating any horizontal interference effects that would have existed with the previous system. The tweeters and raids are standard Allison units (identical to those in the IC-20), with the tweeter/ mid crossover point set at 4 kHz, eliminating both the low-crossover point problem and the test-tone distortion problem. The woofers are two 6.5-inchers, crossed to the midranges at 450 Hz, that are located on the sides of the enclosure near the bottom, reducing the potential mid-bass extension to 90 Hz problem.
This "super-center" system uses two fully second-order AL-125 crossover networks that I modified to work with each three-driver array. Each tweeter, midrange and woofer combination is on a separate circuit (separate crossover), meaning that the speaker system, as with the system it replaced, has to be biamped. There are dual five-way binding posts on the backside of the speaker to allow this, and I do the biamping with the two 130-watt main-channel amps in my Yamaha RX-Z1 receiver that are fed by the dual center-out hookups The crossover modification mainly involved configuring the bass output of the two crossovers to handle just one woofer apiece, because the AL-125 normally has two woofers, one midrange, and one tweeter.
The cabinet is 38 inches tall, a foot wide, and 10 inches deep and weighs in at 57 pounds. The dual networks and the six-driver array give this system plenty of output potential, although each ten-driver IC-20 still has a theoretical edge. All three systems can play considerably louder than I require, fortunately.
The system does not have a perfectly smooth, unequalized room-curve response, but it is considerably flatter in terms of my measurements than the previous center-channel speaker. To flatten things out even more--to within +/- 1 dB from 90 Hz to 8 kHz--I make use of an AudioControl C-131 equalizer. Note that I measure the room response at the 13 foot distance to the main listening couch and not at a close-up distance that would highlight the output in the direct field at the expense of what we normally hear at typical listening distances. All measurements were done with my standard, 20-second averaging technique that involves moving the microphone over a 1 x 1 x 5-foot, box-shaped area at the seated head height as my RTA calculates a cumulative average.
Above 8 kHz I equalize in a gradual rolloff to keep these speakers, normally room-curve flat to 16 kHz, from sounding a bit more brittle than I would like with the classical and baroque material I normally listen to. A rolloff like this works well with home-theater material, too. Below 90 Hz the subwoofer bass output is ramped upward somewhat to compensate for small room acoustics. (The main-channel IC-20 systems are pretty flat in themselves, but I still equalize them to the same tolerances as the center by means of a Rane THX-22 equalizer.)
When I run pink and white noise signals to each of the three channels they sound closer to identical than any other three-front-speaker array I have encountered, including the one previously making use of that earlier center-channel speaker. The combination is superb for both music and home theater, and with music this is particularly true when decoding two channel sources with Dolby ProLogic II. The sense of realistic soundstaging is particularly evident from listening positions away from the centered-up sweet spot.
Second Topic. OK, now describing what I have done with my customized center-channel project is not something that is normally relevant to those who are reading a magazine like this for information about product purchasing and/or what recordings to buy. However, it does relate to what I think speakers should be able to do in typical home-listening situations. It also relates to what Dr. David Rich said about speaker sound and speaker reviewing in issue 106.
For one thing, like the IC-20 systems on the left and right main channels, it makes use of second-order crossover filtration and the performance stress is upon a smooth, broadband power delivery to the room and a very wide, uniform, and broad-bandwidth horizontal radiation pattern. While the close-up, direct-field response is pretty good, the performance stress is not upon a ruler-flat direct-field, first-arrival signal, particularly at all angles in the vertical plane.
But first, here are a few definitions that should make it easier to understand just what he was discussing in his article and also make it easier to understand just what I will be discussing. All of these characteristics exist in a continuum and not in isolation from each other.
Radiation Pattern. In loudspeakers, this involves the polar response characteristics at all frequencies. And, needless to say, radiation pattern into the room will be influenced by the way the speaker system itself is positioned. For the most part, what we hear from loudspeaker systems in typical home-listening rooms, particularly in the horizontal plane, is the radiation pattern and how it interacts with room boundaries and furnishings and not just the power response in isolation or the direct-field, first-arrival-signal response.
Power Response, or sound power. In loudspeakers, the integrated output in all directions, at all audible frequencies. This can be either measured by means of multiple measurements all around a speaker system (such as what Consumer Reports does) or measured by putting a speaker in a sturdy reverberant chamber and measuring the reflected power. One can somewhat gauge power response by measuring at a lengthy distance from the speaker in a typical home-listening room. However, furnishings will absorb enough energy and the walls and other hard surfaces resonate enough or generate standing waves to a large enough extent to make the technique less than perfect.
My room-curve measuring technique, done in a very good room, is mainly influenced by the power response, although the direct-field response and the radiation pattern of the speakers will, depending upon the dispersion characteristics of the speaker, also play contributing roles
Direct Field Dominating. in home, studio, and indoor live-music listening environments, this defines the sound field that exists when the sound coming directly from a source (usually a loudspeaker) is louder than the reflected or reverberated sound bounced off room boundaries or furnishings. Normally, you would have to be very close to the sound source for this to occur at all audible frequencies, although over the top two octaves the direct field can dominate a fair distance out from a typical loudspeaker. In normal rooms, at all frequencies below the treble and at a normal listening distance, its influence pales in comparison to power response and cumulative effect of radiation pattern. Needless to say, the direct-field signal, be it dominating or be it overshadowed by the reverberant-field signals will be the first-arrival signal that helps to stabilize the soundstage and promote precise imaging.
With large panel-type speakers and some horn designs it may be impossible to get out of the dominating direct field in a typical listening room, at least at middle and higher frequencies. With well-designed wide-dispersion speaker designs it may be impossible to sit comfortably close enough to the systems for the direct field to dominate. The direct field can also be influenced by driver positioning, cabinet shape, and crossover, grill, and driver design. While the direct field is best measured anechoically in a chamber or outdoors, it is possible to measure the direct field above the upper bass range indoors if a gated system, such as the Acoustisoft ETF system employed by David Rich in his reviewing work, is used.
Reverberant Field Dominating. In home, studio, and live-music listening environments, this defines the sound field that exists when boundary-reflected sound is louder than the first-arrival sound coming directly from a source, be that source loudspeaker systems or live performers. As such, with speaker systems it is strongly related to power response. Obviously, the strength of the reverberant field will depend upon the listening distance and the room's layout and reflectivity, as well as the dispersion characteristics of the speakers. In most home-listening situations and with decent wide-dispersion speakers the reverberant field may dominate at all but the highest frequencies. With narrower-dispersion speakers, particularly in highly damped rooms, the impact of the reverberant field will be lessened.
Critical Distance. The point at a given distance from a loudspeaker system or other sound source playing in an enclosed space where the direct-field, first-arrival signals coming from the source and the boundary--reflected reverberation generated by that same source are perceived at equal levels. Hence, the "critical distance" defines the boundary between the direct and reverberant fields, and the location of the critical distance will be controlled by both the directivity of the speaker over its operating range and the reflectivity of the room boundaries.
As a rule of thumb, the critical distance is close to a speaker at low frequencies, where wavelengths are long in relation to driver size, and therefore dispersion is wide and able to bounce a lot of sound off of adjacent room boundaries. On the other hand it will normally be further out from a system at higher frequencies, where wavelengths are short in relation to driver size, and therefore dispersion is narrowed and less sound is bounced off of adjacent boundaries.
However, things are more complex than this, because speaker systems nearly always have multiple drivers of different sizes, with those drivers arranged in any number of different configurations. Consequently, the resulting radiation-pattern output can vary over the entire operating range of the speaker as the frequencies being handled change. Even a single driver will have narrowing dispersion as the signals it handles climb in frequency. Consequently, the critical distance will not be a stable point out in front of the speakers, and with virtually all speaker systems it shifts position as the frequency changes. With lesser speaker systems it can shift position substantially.
With narrow-dispersion speakers it may be some distance out from the systems and with wider-dispersing models it may be quite close to the speakers, but with either design it will shift to a degree as the frequencies reproduced change, even in the midrange. Whether the system is a wide-dispersing design or a narrow-dispersing design, the more stabilized the critical distance is over the system's operating range, the better. In my experience, it is easier to stabilize the critical distance over a broad frequency range with uniformly wide-dispersing designs than with narrower-dispersing versions.
End of definitions.
OK, in David Rich's article, "Speaker Testing: an Overview," the author pretty much directly opposed much of what I have said over the past few years about speaker sound and speaker/room interactions. Indeed, at one point he states that when I read his article I "will hit the ceiling." Actually, I did not dare do this, because we recently renovated our house, but I did go outside and read the article again and then had plenty of space to leap into the air at times. His article also runs counter to the speaker performance and speaker measuring overview that David Moran outlined in his review of the B&O BeoLab 5 speaker system in issue 105. One wonders which of these two approaches (David Rich's or David Moran's) is more valid, because each seemed to be backed up by so much objective data.
Then there is the Howard Ferstler approach. For me, in contrast to what David Rich said in his article, the first hurdle a speaker must pass over before moving on to other aspects of its performance is flat power response, and not the flat direct-field response that matters mainly to headphone-detail enthusiasts. In the reverberant field that surrounds most listening positions in normal rooms the power response and not the much weaker direct-field response is what will dominate with musical source material.
OK, now why do I consider the total power output and the related wide-angle (wide angle meaning well beyond 30 or 45 degrees off axis) radiation-pattern output of a speaker to be all that important? There are a number of reasons.
First, what we hear during indoor listening situations in normal rooms is total power: the power response. Yes, I know that we zero in on that first-arrival, direct-field signal for directional clues, and the leading edge of the direct-field transients in the treble can help determine the detail a speaker will deliver and tighten up imaging. However, the fact is that any direct-field, first-arrival signals reaching the ears from a speaker are but a very small percentage of the total sound. Sure, I realize that some critics of the sound-power approach say that if we turn a pair of speakers around and face them towards the sides or even towards the front wall they will sound different from how they sound when facing forward. Certainly, radiation pattern plays a big part in both imaging and spectral balance issues, and adjusting a speaker to different angles influences the shape of the radiation pattern. Facing a speaker towards the front wall, even if you pull it out far enough for it to "breathe" freely, obviously has a huge impact on the radiation pattern out into the room, including the contributions from the boundaries adjacent to that wall. Most speakers are directional (some are more directional than others, and often they are erratically directional over their operating ranges) and the way they propagate their sound into a room is second (and a very close second) only to the impact of the power response of a speaker.
Notwithstanding the impact of aiming speakers in different directions within a room to experience subtle shifts in imaging, tonal balance and initial signal clarity in the treble, all one need do to experience the huge impact of power response and the importance of the reverberant field is to haul their speakers outdoors and listen to the difference between the two environments.
Outdoors, the direct field, first-arrival signal dominates the spectral balance, and the impact of the power response will essentially be nearly zero. (The ground will contribute some reflected energy, obviously.) Consequently, a speaker with a flat, phase-coherent first-arrival signal will deliver an impressive response curve and headphone-like precision. Unfortunately, the system will have a dry, sterile sound out there that simply does not mimic what live music sounds like, at least when performed indoors.
On the other hand, when transported back indoors, the reverberant-field totality of a typical speaker system utterly blows away the spectral-balance contributions of the direct-field, first-arrival signals reaching the ears. Even if the direct field, first-arrival response is only reasonably smooth, if the power response resulting from the total output in all directions is smooth the system will have a subjectively smooth spectral balance - at least in rooms with normal acoustic properties and if auditioned at normal listening distances.
Even if the output beyond 45 degrees off axis is attenuated considerably, as it would be with speakers designed for narrow dispersion and a flat first-arrival signal, the impact of the wide-angular signals will still be considerable. In most typical listening rooms, and at typical listening distances, it does not matter how smooth and flat the direct-field, first-arrival signal is if the speaker has an irregular, choppy response at wide off-axis angles, because the resulting power response will not be smooth.
Indeed, with this 0-45 and 45-90 degree issue still in mind, compare the microscopic amount of energy delivered by the near zero angle of the direct-field, first-arrival signals (one to each ear) to that radiated to and reflected from nearby boundaries involving the total hemisphere. Then, go beyond the hemisphere to the total spherical response of a speaker system in a normally reflective home-listening environment. It is preposterous to think that the direct-field, first-arrival signal has any major impact on the spectral balance of a typical loudspeaker system in typical home-listening situations. On imaging and focus, yes, and perhaps to the initial attack clarity of higher-frequency instruments, but on overall spectral balance--no. Making the direct-field output of a speaker the most important thing when it comes to speaker performance is akin to having the tail wag the dog. Yes, it can be important under some conditions, but it takes a back seat to a few other issues, including power response and the uniformity of the broad-bandwidth radiation pattern and its impact on boundary augmentation.
Now, in his article David Rich states that "flat direct-field response (on and off axis) is a gate that all good speakers must pass through," although he does not mention just how far off axis this standard must go. Certainly, if the broad-bandwidth, direct-field measurements are smooth over a very wide angle around the speaker the total power response will also flatten out and the radiation pattern will be consistent. However, given the multitude of direct-field signals radiating away from a speaker at a multitude of angles and combining to form the power response, it is likely that a system that has a moderately troublesome direct-field, first-arrival response at the listener's ears but still excellent power response will still sound flat and smooth--and good--at normal listening positions. It might not have headphone-like clarity and precision imaging focus, but it will sound more like what we get at a good seating location at a live concert than a super-detailed, super-flat first-arrival, highly directional speaker that lets you hear the performers breathing and score pages being turned.
Actually, I get the impression that response smoothness at extreme angles is not something that particularly concerns David Rich. Indeed, he indicates in one of the reviews following his introductory essay that a tweeter design that narrows the dispersion angle will possibly result in a sound field that "spreads across the space between the speakers and does not just hang around the speakers." According to him, it will do this, because the narrow dispersion "limits early sidewall reflections." However, I have not found this to be the case at all, and I would guess that I have auditioned more speakers with wide-bandwidth, super-wide and super-smooth radiation patterns than has David.
Indeed, a system that generates strong and smooth--and wide bandwidth--early sidewall reflections will have a more uniform soundstage spread and sense of concert-hall space when listening from an off-axis listening location than a speaker that has a more narrowly aimed output. David Rich does admit that wide dispersion in the treble will add a degree of "so-called 'air'" to the sound, with narrower dispersion designs often sounding a bit dry. Unfortunately, he does not make a value judgement when it comes to these diverging approaches. What is better: a dry, focused sound or a wet, airy sound? Obviously, some recordings will benefit more from one characteristic than the other, but there is no hard and fast rule in any case.
While he does provide us with some sound-power curves in the speaker reviews that follow his introductory article (run by Harman International, actually), most of his serious measurements are confined to 45 degrees off axis or less. And a lot of copy is spent discussing the direct-field advantages of super-steep crossover slopes and the disadvantages of driver-response overlaps, particularly in the vertical axis, even though the importance of such things is greatly reduced when the reverberant field dominates. He also gives us a clue as to why he believes as he does about the sound of speakers and the importance of the direct field: he indicates that his reference standard for high fidelity sound is his pair of Sennheiser HD-600 headphones.
OK, I will concede that headphones deliver the genuine goods if someone wants to monitor completely within the direct field and get maximum, old-fashioned "Hi-Fi" impact. You hear things with headphones that you would never hear with speakers. However, you also hear things with them that you would never hear at live performances, unless you are sitting in the front row at the hall or perhaps conducting the orchestra.
Clearly, having a speaker with a smooth, coherent direct-field, first-arrival signal, at least when listened to up close, and with perhaps some room padding treatment installed on the walls, will allow one to hear things with borderline headphone like clarity. The question is: with typical good recordings do speakers that work this way sound significantly more realistic in home listening situations than speakers that stress flat power response at the possible expense of a ruler flat direct-field, first-arrival signal?
Ironically, in contrast to other reviewers who are enamored of flat direct-field sound from speakers, David Rich is not particularly caught up in the super-imaging issue, where a near-perfect direct-field signal would be important. He states in the article that "pinpoint imaging does not exist in a real hall even if you sit up close with your eyes closed." No, for him the issue is spectral balance, detail, and smooth sound and he sees the direct field as vitally important in that realm. I agree completely with him on the issue of tight imaging, but I disagree when it comes to the importance of the direct-field signal.
Obviously, someone who prefers headphone sound is going to want speakers to be major-grade performers in terms of direct-field smoothness at the listening position, with minimal overlap between driver outputs, both horizontally and vertically. Probably, they will also sit rather close to their speakers, will have those speakers sited some distance from room boundaries (to minimize short-delay reflections), and will possibly have their listening room treated to some extent to minimize spurious off-axis reflections that contribute to the reverberant field.
If the room is untreated and the listener is not close enough to the speakers, they are going to mostly be on the far side of the "critical distance" boundary in front of the systems that determines where the direct-field dominance ends and the reverberant field dominance begins. Get too far out and the reverberant field strength will swamp that of the direct field, at least from the midrange to the lower treble. If the tweeters and midranges are wide-dispersion jobs (the ones I use in my main system are, while those used in the main speakers of my also very fine middle system are not) even the treble and upper midrange may be in the reverberant field fairly close to the speakers.
OK, if close-up listening and/or room-padding treatments are their bag, more power to their preferences. In a Stereo Review magazine article I wrote years ago (October, 1990, with related articles in December and April of 1995) I acknowledged that both approaches (a broad-bandwidth reverberant field dominance or a broad-bandwidth direct-field dominance) were workable and subjectively valid. This was the case, provided that the speakers could deliver either smooth direct-field dominated or smooth reverberant-field dominated signals to the listening position and the critical distance was reasonably stable. One approach might work fine with some recordings and the other might work fine with other recordings--with either approach sometimes working just fine (but differently) with still other recordings. Both points of view have validity, and I think it is rather unfair to set up standards that laud one and dismiss the other. Yep, engineering specifics notwithstanding, taste does play a part in speaker performance, as does speaker design, speaker placement, room acoustics, and the recordings one listens to.
Finally, at the conclusion of one of the speaker reviews that follow his introductory article, David Rich stated that "under no circumstances should you add a subwoofer [to the speaker combination he is reviewing], which is only going to gum up the all-important upper bass that these speakers produce so well."
OK, I would certainly agree with this statement if the subwoofer was crossed over within or above the "upper bass" range. However, good subs, properly configured, are crossed over considerably lower in frequency than that. When this is done a good subwoofer will have the potential to improve the performance of just about any satellite speaker package, if only because it relieves the bass/midrange or bass drivers from handling deep-bass signals. Provided a low-distortion sub is crossed over high enough for the satellites to still blend smoothly (which is admittedly problematic with really tiny satellites) and yet low enough to eliminate localization clues and not muddy up the upper bass, there is no subjective or technical reason why a good sub/sat system cannot perform admirably with musical source materials.
KWN comments: Normally I prefer to stay out of such discussions, but in this instance I thought it might be beneficial for our readers to point out that this is not necessarily a case of one side is right and the other is wrong. There are a lot of variables in speaker design, and even more variables in how different speaker designs are perceived to perform. HF has long been and advocate of the Roy Allison school of flat power response, while DR is likely to cite the more recent work of folks such as Sean Olive. Olive, for example, points out in a recent paper that from many trained listeners, speakers with flat power response were reported as sounding overly bright. Given the variations in taste that exist among our readers, not to mention the variation in recordings and rooms and speaker placement, I think you can see that it is not a back-and white issue being raised here.
The important thing is for readers to note the underlying principles that both DR and HF have appealed to in their explanations of speaker performance, and to realize that for both HF and DR, the room is an important factor. HF does much of his listening in a large room, while DR listens in a much smaller room. This is going to affect how they view speaker design.
Another point worth noting is that both HF and DR are committed to the importance of good frequency-response performance in loudspeaker design. The may disagree somewhat in terms of just how speaker should be measured, but neither is ready to throw good frequency response out the window.
My advice, then, would be to learn from both these gentlemen, as both have a lot of good advice to offer. If either of them offers high praise for a loudspeaker, then chances are, it is a good product, well worth an audition.
|Printer friendly Cite/link Email Feedback|
|Article Type:||Product/service evaluation|
|Date:||Apr 1, 2006|
|Previous Article:||Make my radio HD, please!|
|Next Article:||Surveying the soundscape.|