By the Staff of The Audio Critic

Part I: Comprising seven big ones, four little ones, three medium sized jobs, plus a decent subwoofer, which should be enough to keep everyone arguing until Part II.

There are good reasons why speakers are the most controversial of audio components.

First of all, they are the least accurate, introducing much greater discrepancies between input and output than amplifiers or tape recorders or even phono cartridges. That makes most comparisons highly dilemmatic, as they are essentially between varying degrees of

"bad." (Who was the nicer guy, Adolf Eichmann or Charles Manson? You're wrong!) This basic judgmental problem never goes away; it's there whether one is comparing the Beveridge and Koss electrostatics or the house-brand specials of a hi-fi chain store. Name your preference and someone will immediately tell you what's wrong with it. And he'll be right.

No speaker, at any price, is without faults.

Another cause of unending controversy is that a lot of audio enthusiasts with strong opinions have never heard, let alone lived with, the best available speaker systems, which are usually made in extremely limited quantities and demonstrated in very few places. As a result, the typical arguments tend to be parochial, as they are engaged in without awareness of what is currently possible. Amazingly, even speaker manufacturers suffer from this limitation; you'd expect their sound rooms to bulge with the latest and greatest multi-kilo-buck speakers for reference purposes, but the truth is that most of them (rich or poor!) listen exclusively to their own products. Their dealers are usually way ahead of them in listening experience.

The most profound source of confusion, however, is the quality of the signal fed to the speakers in nearly all comparisons. A record made with peaky Neumann microphones is placed on a turntable with insufficient feed back isolation and a motor board that drums. A cartridge mounted conventionally, with un necessarily large tracking error, in a tone arm with sloppy bearings (cf. our cartridge/arm/ turntable article in this issue) is lowered on

the record. The signal is applied to the speakers through an amplification system with TIM, in sufficient voltage swing, power supply limitations, etc., etc. The speakers are then found "crisp" or ''nasal" or "honky" or "airy" or whatever. And these observations are then passed on to audiophile friends or even published in underground audio reviews. If at least the signal had the same faults in each listening test! But, needless to say, its quality varies completely at random, and so do the conclusions.

How we tested them.

What we said about A-B listening tests in connection with subwoofers in our second issue (March/April 1977, page 30) applies equally to full-range speaker systems. The audible differences are so gross, between any two models at any volume level, that A-B switching becomes a pious exercise in methodology yielding no additional information. When you can find two really good speakers that have to be carefully A-B-ed to establish which sounds more ac curate, you'll know the art has advanced by a whole order of magnitude.

On the other hand, one-by-one listening evaluations require absolutely unquestionable signal quality, and we went all out to obtain it.

Only the finalists of our preamp and power amp surveys were used (incidentally, the new Hegeman HPR /CU preamp has emerged as a very serious contender for reference status, al though we won't be ready to rate it against the Mark Levinson, Rappaport and others until the next issue); our current reference cartridge, the Denon DL-103S, was religiously aligned for optimum tracking both laterally and vertically; we eschewed all records, no matter how dear to our heart musically, unless they were obviously recorded with flat microphones, like the Mark Levinson Acoustic Recordings; and, to establish an ultimate reference point, we unleashed our new, highly modified Stellavox 'Stella master' tape deck, playing direct 15-ips copies of selected original master tapes. (Please don't write us to ask how we buy, beg, borrow or steal these.) In other words, we were listening to the speakers, not to signal distortions ahead of the speaker terminals.

After extensive listening, we measured.

We made a point of starting our laboratory tests only after we had formed a reasonably firm opinion of how a speaker sounded and what was audibly right or wrong with its performance. We didn't want to expose ourselves to the temptation of subjectively aha-ing and you-seeing from our listening chair what we had already established objectively through measurement. Certainly not the first time we heard a speaker. All the initial aha's took place in the laboratory, as they should-and there were plenty of them. Unlike, say, preamplifiers, loudspeakers haven't evolved yet to the point where they can hide their faults from measuring instruments. When a speaker doesn't sound right, it doesn't measure right-and since not a single one sounds quite right, not one measures quite right. But, gratifyingly, the ones that sound better measure better. Needless to say, more than one or two kinds of measurement are necessary, and the results are seldom cut-and dried, being generally the reflection of deliberate design trade-offs and therefore subject to interpretation. Even so, we're convinced that if every speaker manufacturer routinely per formed our simple tests (as we know for a fact nine out of ten don't), speaker design would be a far less haphazard affair today and the art would advance more sure-footedly.

We must emphasize, however, that our main concern in the laboratory was to make sure we hadn't made a mistake in our listening evaluations-that what we're reporting here is based on some sort of reality rather than on irrelevant artifacts of our component chain or on the aural idiosyncrasies of our staff members. For that reason, the overall thrust of our measurements was qualitative rather than quantitative; in the majority of cases we went only as far as it seemed necessary in order to authenticate our conclusions and recommendations-not to determine by how many dB or Hz or msec a manufacturer ought to change his design parameters to end up with a better speaker. (We try to give our subscribers useful advice and reliable explanations, but for $28 a year we're not going to give any company a product research and development program.) After measuring, we listened again, this time of course "corrupted" in our subjectivity by our newly gained knowledge. And then, in some cases, we went back to the laboratory once more, just to double-check certain observations and correlations. The remarkable thing about this process is that it's synergistic; one becomes a more discerning listener after having objectively verified in the laboratory some purely subjective impressions, and one measures with greater astuteness and more concrete results after having aurally zeroed in on certain sonic peculiarities. Pretty soon we began to wonder how anyone could possibly form a serious opinion about a speaker without going through these steps-forgetting that there was a time when we didn't quite do it this way our selves. (Who says a critic must be born perfect and never undergo any development?) The lab measurements.

We measured each speaker in both the frequency domain and the time domain. We're convinced, without any reservation, that one without the other yields an incomplete picture and can't be satisfactorily correlated with the overall subjective impression made by a speaker.

We do lean, however, toward the time domain as the more important of the two-otherwise why should a live singer or a live piano heard through a half-open window all the way around the corner still sound unmistakably live instead of canned? The frequency response is certainly shot to hell that way; only the time response retains its integrity, since all components of the sonic information are still traversing the air path at the same speed. (When you get right down to it, there's no such thing as "frequency" - identical sine-wave cycles per unit time-in the real world: it's an artifact of the human mind.

But there is time all right, and nothing remains the same from instant to instant.) Enough of this philosophizing. Here's what we did: We measured the 'nearfield’ frequency response of each driver in each speaker. This corresponds very closely to anechoic measurement, as already explained in the subwoofer article in our second issue. We also took the overall frequency response curve of each complete system and, in some cases, of pairs or sub groups of drivers within the system. Both linear and log sweeps were used in these measurements, and both axial and power response were investigated.

Whenever it appeared that single-frequency distortion would be relevant to our evaluation, we checked the output of individual drivers for THD at the frequencies in question.

We ran quite extensive tone-burst tests on each system, at many different frequencies, with both sine-wave and square-wave bursts.

Perhaps most significantly, we tested each system with single pulses (call them blips or ticks, if you like) ranging in width from approximately 1 msec to 0.1 msec, spaced so far apart that room reflections didn't enter into the picture. Both the oscilloscopic trace and the audible quality of such pulses clearly reveal time-response information not easily obtain able any other way. The differences from speaker to speaker are absolutely startling on this test: some will reproduce a very adequate copy of the input pulse, with no garbage in the silent interval that precedes the next pulse; others will generate an output totally unrelated to the input.

In addition to these basic tests, we also listened to (and occasionally spectrum-analyzed) white noise and pink noise through the speakers, checked for structural faults, investigated impedance characteristics when we suspected significant peculiarities (but only then), and performed other little peripheral testing chores.

We must repeat: our emphasis was on insight rather than on the accumulation of raw data.

The laboratory instruments we used were the Bruel & Kjaer 4133 calibrated microphone, the Hewlett-Packard 3580A spectrum analyzer, the Hewlett-Packard 1740A oscilloscope, the Wavetek 185 function generator, the Sound Technology 1700B distortion analyzer (with built-in oscillator), the General Radio 1382 random-noise generator, and the Hewlett Packard 3435A digital multimeter.

What about our usual sweeping generalizations on design criteria? Designing an electroacoustic transducer isn't at all like designing a preamp or a tuner.

No one has ever come even close to theoretical perfection in a full-range speaker system, so it would be wrong to assert categorically that successful design must proceed in this or that particular direction. Somebody would promptly go the opposite way and probably do just as well. We therefore refuse to take sides on electrostatic vs. electrodynamic drivers, crossover less vs. multiple-crossover systems, sealed vs.

vented boxes, or any other design approach that can be either 95% right or 95% wrong, depending on how it's executed. We will take sides when it comes to the basic laws of nature (say, a claim of deep, loud bass out of a 1/4-cubic-foot mini-system) or obviously wrongheaded engineering (like an incorrectly tuned vented enclosure, for example).

We rashly promised in our second issue that we'd go more deeply into crossover design considerations in this survey, but the subject happens to be a bottomless pit. No sooner do we come to the conclusion that high-order net works with their very steep slopes can't possibly be any good on account of their nightmarish time-delay characteristics, one of our consul tants comes in with mathematical proof that with certain kinds of time compensation they can be made to pass a perfect square wave. Of course, he adds, no commercial design is likely to take that route, because of its ridiculous complexity and high cost! And so it goes.

Since each speaker design turns out to be a law unto itself, our only choice is to make our generalizations, such as they are, under the individual product headings. So-we'll let the show begin.

Acoustat X Acoustat Corporation, 4020 North 29th Avenue, Holly wood, FL 33020. Acoustat X full-range direct-drive electrostatic speaker system, $1995 the pair (with built-in power amplifiers). Five-year warranty, excluding tubes; customer pays all freight. Tested #491 and #492, on loan from manufacturer.

Once again, the first piece of equipment in our alphabetical listing poses a special problem.

We approach this review of the Acoustat X with great trepidation. The subject seems to be booby-trapped for the reviewer.

Why? Because, long after the completion of our tests, it remains unclear to us whether our samples were absolutely typical. Objectively, we must conclude that they were indeed typical and that our findings are applicable across the board. Yet, there's a persistent mystique to these speakers out there in audio freak country, suggesting that maybe (just maybe) occasional super samples appear in circulation. The manufacturer professes to have no knowledge of such variability. Let's marshal the evidence:

Last winter, we heard the Acoustat X at a nearby store and were quite favorably impressed, although we did notice some peculiarities. We briefly reported so in our very first issue. In retrospect it seems that this was a better sound than we were subsequently able to obtain from our test samples, but then retrospect is a notoriously poor witness. We have since quizzed every audio enthusiast known to us who has had any experience with the Acoustat X, and their reports are wildly contradictory.

A number of aficionados, including professionals with absolutely impeccable credentials, are completely sold on the Acoustat. State of The Art, they say. Others, equally expert-and these, unfortunately, happen to be the ones whose ears we have always trusted and who listen the way we do-have a much lower opinion of the speaker. The clincher came at the Summer CES in Chicago. The pair demonstrated in the Acoustat exhibit (presumably under the watchful eye of the manufacturer) sounded exactly-but exactly-like our test samples. As a matter of fact, a well-known and highly respected European record producer, whom we had met earlier, drew us aside as we were leaving the room and whispered: "Is this the famous Acoustat electrostatic? Does it sound right to you?" We rest our case.

To get down to business, the Acoustat X is a four-foot high dipole radiator consisting of three long, narrow, side-by-side panels, slightly angled with respect to one another, the middle one a bit narrower than the outside ones. The panels are full-range electrostatic transducers, driven directly off the plates of the 6HBS output tubes of a matching hybrid power amplifier, which screws into the base of the speaker frame (or call it cabinet). Left and right channels are completely identical. On paper, the whole thing is a purist's dream: all the advantages of electrostatic (i.e., force-over-area) drive; no cross overs, hence no phase and time-delay problems from that source; no transformers in the signal path, as in most other electrostatics; no signal processing stages between the preamp output and the ear that could be further eliminated.

And yet . . .

All right, first the good news. The Acoustat X does have the unmistakable midrange integrity of a crossover-less electrostatic design. As long as the program material stays in a relatively narrow frequency and dynamic range (like, say, a singer with acoustic guitar), the sound is quite natural and likable. That alone may possibly account for all the favorable reactions. (Including ours in the dealer's show room.) But feed the speaker some wide-range symphonic material, with banging timpani, clashing cymbals and soaring strings, or some heavily produced rock, and the naturalness is gone-even at quite reasonable volume levels.

The speaker begins to sound strained and steely on top, bumpy and muffled below. The free breathing ease and openness that characterize the reproduction of less complex material are replaced by a disturbingly canned quality. Vigorously played strings sound "electronic" even when they aren't massed, as in a trio or quartet.

So does the piano.

We believe we know the reason for this behavior. Our tone-burst tests revealed substantial ringing at a number of frequencies, especially around 1 kHz but to some degree also around 4 kHz and several other spots. The energy peaks of complex, dynamic program material seem to excite these problem areas to the point where the ringing becomes distinctly audible and the sound quality suffers.

The most plausible cause of the trouble would appear to be standing waves in the large panels, especially since the ringing is definitely lower in amplitude near the clamped edges.

Interestingly enough, single pulses are also very difficult to reproduce accurately through the Acoustat X. A crossover-less, force-over area system could be expected to have all the advantages when it comes to this time-smear test (cf. the Beveridge below), but the Acoustat panels fail to render a good likeness of the pulse shape and are altogether unable to repro duce certain pulse widths (such as, for example, 0.35 msec). Whether this is symptomatic of the same problem as the ringing or unrelated to it we aren't ready to assert; it was just about at this point that we ran out of R and D time for the Acoustat Corporation. In any event, the Acoustat X lacks the ultimate definition, focus, and delicacy of inner detail that characterize a speaker with outstanding pulse response.

In the frequency domain the speaker is quite good, though not spectacular; overall power response is reasonably flat; there are, however, some very nasty lobes in the polar response. As a matter of fact, vertical beaming is so severe that it's hopeless to listen to the speakers standing up: your ears must be approximately level with the midpoint of the panels, otherwise you don't hear the highs. We could live with that, though, if the speaker had no worse faults.

One of these worse faults is in the deep bass: a great big bump in the response at 35 Hz that creates utter chaos whenever it's excited.

The speaker actually sizzles when the signal generator sweeps through this bump, and organ pedal notes or bass-drum whacks anywhere in that vicinity overload the speaker to the point of offensive flatulence. It's even possible that the higher harmonics of this powerful resonance are responsible for that bumpy, muffled quality we discerned in the upper bass and lower mid range on complex program material. In addition, both the speaker frame and the amplifier chassis buzz at various frequencies. It's rather a mess.

Much has been made of the sensitivity of the Acoustat X to room placement, of the touchiness of its high-frequency balance control and even of its amplifier gain control, of the pros and cons of the built-in equalizer network that counteracts the boundary effect of the rear wall (it can be disabled by padding with a 470K-ohm resistor), of the desirability of raising the speakers off the floor, and so forth and so on. All we can tell you is: that's not where it's at. All these things make a difference, but not the Big Difference. We experimented with them all, staying in close touch with the manufacturer (who even sent us "new and improved" IC's for the front end of the amplifier-they were indeed better). We ended up using the speakers exactly as the instructions specify: 2 to 3 feet out from a fairly live wall, looking into the longer dimension of the room. That was by far the best arrangement, but they still didn't sound like reference speakers for the purist.

It must be remembered, of course, that the Acoustat X isn't really expensive-not when you consider that the power amp is included.

That puts it almost in the same price category as the Dahlquist DQ-10; certainly not a whole lot higher. On the other hand, there's always the possibility that it's the amplifier that's causing some of the problems we observed. We doubt it, though. Since neither the speaker nor the amplifier alone is compatible with other equipment, the question is academic, and we didn't investigate it further.

Oh yes. We forgot. Both amplifiers eventually conked out in the course of our testing.

One in the middle of some loud music, the other in the middle of a tone burst. Neither responded to elementary first aid. But by then our conclusions were firm. And will remain so unless we're presented with totally new evidence to the contrary.

Beveridge Model 2SW Harold Beveridge Inc., 422 North Milpas Street, Santa Barbara, CA 93103. Beveridge Cylindrical Sound Sys tem, Model 2SW, 85200 the pair (with subwoofers, plug-in power amplifiers and control module). Virtually unlimited guarantee ('you must be absolutely satisfied").

Tested #233 and #234, on loan from manufacturer.

Let's say it right at the beginning. This speaker/amplifier system, despite its undeniable limitations, comes closer to our ideal of sonic accuracy than any other we've ever heard, with the possible exception of Mark Levinson's stratospherically priced HQD System (which isn't really a single product but an ad hoc blend of four different brands and which we haven't tested yet under our own roof). After that statement has duly sunk in, we must add that some audio perfectionists will never settle for a sys tem that can't play louder and has no more head room than the Beveridge. Not at this price. To us, however, its specific virtues are irresistible.

The SPL /headroom/dynamic-range problem was of course the very reason why Beveridge didn't stick with the original Model 2, which was "purer" conceptually: a monolithic, crossover-less, full-range electrostatic system, with matching direct-drive high-voltage amplifier (a la Acoustat X but vastly more sophisticated, as we shall see). The Model 2SW, by incorporating a 70-Hz crossover (6 dB per octave and built right into the power amp) and letting a conventional electrodynamic subwoofer handle the lower bass, can play considerably louder, since the frequency division permits the electro static main unit to be optimized for the narrower range. Even so, forget about any SPL above 100 dB or thereabouts in a fair-sized room. The Beveridge sounds so good when it doesn't have to strain that it's a crime to strain it. (Besides, the fuse will blow.) Why does it sound so good? There are two main reasons, in our opinion. One is that it's probably the only speaker that's a true line source and lays a sound field into the room accordingly. (The Infinity "Quantum Line Source," for example, isn't and doesn't.) The other is its incredibly good impulse response, indicating extremely low time-delay distortion, which audibly outweighs whatever minor deficiencies the speaker may have in the frequency domain.

A quick description of the design: A 6-by-1 foot electrostatic transducer is front-loaded by a 6-foot high "acoustic lens" terminating in a relatively narrow aperture. The rear radiation is blocked. The lens, which is in effect a system of vertical guiding walls, bends the planar wave produced by the electrostatic panel into a 6-foot high 180° cylindrical wave front, invariant with frequency, in accordance with the geometry of a line source. The transducer, the lens, and the baffle structure for the back wave are all housed in a huge, upright, coffin-like structure that plugs directly (through contact pins in its bottom) into a metal base that houses the amplifier.

The system is best set up with the apertures of the left and right units facing each other across the room. The two subwoofers (large but manageable) plug into their respective amplifiers via ordinary speaker wire and banana plugs. A 30-foot shielded cable emerges from each amplifier to be plugged into your preamp. A separate control module is also supplied with the system, with provisions to trim the frequency response within a few dB ("spectrum slope' and "'bass environment") and to vary the stereo spread ("lateral control'"). We found that this control unit introduces an ever-so-slight (call it infinitesimal) degree of veiling, so we unplugged it, especially since our listening room characteristics happened to suit the naked speakers very well. On the test bench, the control unit gave no evidence of any type of distortion; indeed, it's very well designed.

The sound of the Beveridge Model 2SW is such that, at a reasonable SPL and with superior program material, most listeners go '"'yechh" when any other speaker is switched in for com parison. Even the good ones in our survey. The Beveridge is far superior in transparency, mid range and (especially) top-end openness, freedom from nasality and vowel-type colorations, depth perspective, attack and release, and overall delineation of inner detail. Further more, that cylindrical wave front really does what it's supposed to; uniform distribution of the direct-path "first-arrival" sound results in absolutely unvarying balance and perspective anywhere in the room-near or far, left or right, sitting or standing, or even right between the speakers. Since the SPL from a line source varies inversely with the distance, not the square of the distance as in the case of conventional speakers, the uniformity of the sound field is even more startling. Stereo imaging is also excellent and quite independent of listener position, but be prepared for differences in comparison with good point-source speakers, such as the Rogers. The fact is that the exact complementary playback geometry for a particular microphone setup is always conjectural and therefore highly arguable, so that the characteristic wall-to-wall stereo image created by the Beveridge is as valid as any and at least never smeared by time delay.

What all this adds up to is a strikingly original design concept executed with a high degree of competence. Not since the Lincoln Walsh speaker has there been anything com parable in sheer creative thinking-and the actual implementation of the Beveridge idea is of course far superior (cf. the Ohm F below).

It's interesting to note that Harold Beveridge, who is responsible for both the theory and the practice, never had anything to do with the audio industry, having spent his entire engineering career in high-level national defense jobs.

It seems that hanging out with Bill Johnson or Joe Grado or Dick Sequerra isn't the only way to learn about the frontiers of audio.

We must now point out what's wrong with the speaker. For one thing, the woofer is only pretty good instead of great, which is what the Beveridge electrostatic would deserve. The 12 inch woofer in its sealed enclosure appears to have a system resonance of 36 Hz, since at 40 Hz its unequalized amplitude response is up 5 dB, indicating a Q of 1.7. The response drops back to the 0 dB level at 30 Hz. This is not the way to extend response down to 30 Hz, in our opinion; in fact, it doesn't sound like 30 Hz response but rather like bumped-up 40 Hz response, which is of course what it really is. For example, the open E string of the double bass, vigorously plucked, makes the woofer go "boom" instead of "thunk”--and that's almost exactly 40 Hz. It wouldn't happen with a Q of 0.707. Harmonic distortion is again pretty low but not fantastic; the original Acoustic Research woofer of 1952 was better in that respect. (Ed Villchur, where are you now that we really need you?) There's also quite a nasty resonance at 76 Hz in the large (electrostatic) enclosure. It's a fairly broad peak of about 8 dB amplitude and you can hear it all the time if you listen carefully.

Editor's Note: At press time we're told that Harold Beveridge has identified the problem as due to insufficient damping material behind the electrostatic panel, has cured it completely (to 0 dB), and will make the modification in our review samples at the earliest opportunity. Not in time, however, for this already late issue.

The frequency response of the complete system is otherwise very smooth to 12.5 kHz, displays some slight perturbations in the 13-14 kHz region (probably due to standing waves be tween the lens elements), peaks very gently at 16 kHz and then dies quite fast. The far-field integrated response, however, smooths out these minor, and to our ears inaudible, irregularities, so that the picture is one of a very gradual downward slope from 12.5 to 20 kHz, without significant peaks. Overall, we have no quarrel with the frequency response of the Beveridge from, say, 100 Hz on up.

In any event, it's the time response of the speaker that puts it in a class by itself. It reproduces single pulses of just about any reasonable width with startling accuracy. And you don't have to hunt for a "sweet spot" to read a good pulse, as you do with even the best conventional speakers; just point the measuring microphone in the general direction of the speaker-high, low, sideways, over your shoulder while talking on the telephone-and the pulse will be a good replica of the generator output. Again we must say-that cylindrical wave front sure does its job.

One more thing. We haven't heard the original, simon-pure Beveridge Model 2, so we can't comment on the similarities and differences. According to Harold Beveridge, how ever, the overall gain in headroom is almost 10 dB with the Model 2SW, which would totally disqualify its predecessor in our estimation, since even the Model 2SW is just adequate in that respect. In fact the I-think-I-just-heard-it clip crowd will find it rather frustrating, as we've already said.

To us, the speaker remains a joy. It's the only one in this survey that can occasionally make us forget that we're listening to reproduced music. Now and then, it almost sounds live. And that makes it our reference speaker, until somebody sends us something better.

Braun 'Output C Adcom, 114 East 32nd Street, New York, NY 10016.

Braun 'Output C' miniature speaker system, $224 the pair. Tested samples on loan from distributor.

Imagine a speaker system no bigger than one of the heftier twin-lens reflex cameras and in black metal, too. Imagine it with sound that's quite respectable for any speaker, regardless of size: just a notch or two short of top-notch and minus the deep bass. You now have a fairly good picture of the Braun Output C. Amazing and lots of fun.

We wouldn't hesitate to recommend Braun's small wonder to anyone who is really short of space and doesn't even have room for anything like the little Rogers LS3/5A (which is far superior but about three times as big and twice as expensive). The Output C is also the perfect gift "for the audiophile who has every thing." A great toy.

Don't let anyone tell you, though, that these diminutive speakers plus a commode-type subwoofer with summed channels constitute an

"invisible" system approaching SOTA quality, as has been suggested in some quarters. No way. The Output C has some definite colorations even in the range where its tiny woofer and dome tweeter are reasonably flat. These colorations, however, aren't particularly irritating musically; the overall sound of a pair of Output C's used as a full-range stereo system is open, smooth and listenable beyond all expectations. Even the bass is subjectively accept able, although it's mostly fake of course, achieved with bumped-up response at the sys tem resonance of 200 Hz rather than with genuine low-frequency output.

The most noticeable coloration is due to the tweeter, which rings rather badly and doesn't allow the system to reproduce pulses accurately, especially as the pulses get narrower. The result is just the slightest zippiness and a lack of the utter transparency you can expect with superior impulse response (cf. the Rogers).

Even so, we have nothing but admiration for the Braun Output C. The very fact of its existence is a sign of progress in audio engineering.

Braun L200

Adcom, 114 East 32nd Street, New York, NY 10016.

Braun L200 small bookshelf speaker system, $260 the pair. Tested samples on loan from distributor.

These neat little Braun speakers are some what anticlimactic after the Output C's; they're about twice as large (i.e., still exceedingly small) but not much better sonically, so the amazement factor is considerably smaller.

The differences: black plastic rather than metal enclosure, larger woofer of course, same type of dome tweeter but not quite the same diaphragm material. The system resonance is at 175 Hz instead of 200, bumped up the same way (the Q is approximately 1.5). The tweeter still rings, pulse reproduction still deteriorates as one goes from wider pulses to narrower ones, but there's some improvement in these areas.

The overall sound is perhaps somewhat rounder and more authoritative, but not very different.

The slight coloration is still there.

If it were possible to avoid comparing the L200 with its little brother, we'd be impressed: under the circumstances, it's the Output C that stands out as the more remarkable product.

Canton LE 400

Adcom, 114 East 32nd Street, New York, NY 10016.

Canton LE 400 bookshelf speaker system, $350 the pair.

Five-year warranty; manufacturer pays return freight.

Tested #466872 and #466888, on loan from distributor.

Through an alphabetical coincidence, the three German sealed-box mini-systems in this survey follow one another without interruption.

The Canton LE 400 is by far the largest: about half a cubic foot in volume. You'd expect it to have better bass than the two little Brauns and it does; the system resonance is at 110 Hz, where the response is up 6 dB, meaning that the Q is approximately 2, which is definitely on the loosey-goosey side. The corner frequency of the bass roll-off is 74 Hz, as near as we can tell.

The manufacturer claims response down to 35 Hz, which of course is nonsense.

There's actually a three-way system shoe horned into that tight little box, with cross overs at 750 Hz and 2600 Hz. Frequency response is extremely flat from about 700 Hz on up. The star performer is the tweeter, which has astonishingly flat response on axis to well above 20 kHz and even at 45° off axis goes out past 15 kHz. That's superb amplitude response and dispersion in any league. It's also the most likely reason why the speaker system sounds impressively open and "present" on first hearing.

We say first hearing because, after a while, the LE 400 begins to sound over-bright, "electronic" and fatiguing. The culprit is again ringing, easily identified by tone bursts, especially at 2 kHz and 4 kHz. Pulse reproduction is also surprisingly poor.

If you have a chance, though, listen to this speaker. It's the least expensive we've ever heard that, if only for a minute or two, has the very special sound of the top systems with fast tweeters. Then it's all over, but we're willing to bet that the sale is often made by then. Cute little rascal.

Cizek Model #1

Cizek Audio Systems, Inc., 15 Stevens Street, Andover, MA 01810. Model #1 acoustic-suspension loudspeaker, $396 the pair. Five-year warranty; manufacturer pays all freight. Tested #2054 and #2065, on loan from manufacturer.

This is one of the medium-sized units in our survey: a solidly built acoustic-suspension system of the largest bookshelf type (almost too large for a bookshelf), with 10-inch woofer and l-inch dome tweeter. Despite its conventional format, it comes with a special audiophile reputation, since the company is technically oriented and the speaker is more sophisticated in engineering as well as in audible performance than such speakers generally are.

Its best feature is its bass response; the system resonance is at 39 Hz (the specs say 38 Hz, which is accurate enough for us), and the damping is absolutely correct. In fact, at the flick of a switch on the front panel, you can have a Q of 0.6 for dead-flat bass that drops to -4 dB at resonance, or you can choose a Q of I (just a little looser but still well controlled) for 0 dB response at resonance and the slightest ripple just above. These people know their P's and Q's; they also know better than to try to squeeze more out of a given woofer or box size than Mother Nature permits. As a result their bass is close to state of-the-art within the physical limitations of the system-solid, well-defined, musical. It's so easy when you do your homework.

The most highly touted part of the speaker is the crossover network, which is made to see a resistive rather than a reactive load by means of additional impedance compensating net works across each driver. Cizek claims that there's no other way to achieve absolutely flat response in the crossover region, which is a sensitive part of the spectrum in this case, the nominal crossover frequency being 1.5 kHz. An other sophisticated touch is the combination of two high-frequency controls on the front panel (right next to the Q control), one for contour, the other for overall level. Their net effect is rather subtle and, in our opinion, something of a red herring in view of more serious matters in the high-frequency area that require our attention. The least of these is a 5 dB peak at 16 kHz with a precipitous drop thereafter, which is what we blamed initially for imparting a somewhat irritating zing to the otherwise smooth and classy sound of the Cizek. That wasn't it, though. It turned out to be much heavier stuff.

Exploring the speaker with tone bursts, we came upon an astonishing amount of ringing at 3 kHz. It seems that the tweeter reproduces the burst quite accurately, after which the much slower woofer, still quite active at that frequency, chimes in with another complete burst of almost the same amplitude. Two bursts for the price of one: you feed in, say, a four-cycle burst and out comes an eight-cycle burst with just the tiniest cleft in the middle.

Needless to say, at 3 kHz, which is right where the ear tolerates absolutely no nonsense, this spurious doubling of energy sounds like excessive brightness or glare. The ringing continues higher and lower, too, but 3 kHz is its head quarters.

Our pulse tests confirmed the lack of synchronism between woofer and tweeter to be the

speaker's chief weakness. The tweeter is much too fast for the woofer-or, if you prefer, the woofer is too slow for the tweeter. First the tweeter blips, then the woofer; you can actually hear the double blip, unless you put the woofer so much closer to your ear that the speaker, laid on its side, faces inward instead of out toward the listening area. The microphone, too, reads the best pulse from this extreme off-axis position. A time-compensating network could probably do wonders for the Cizek.

Where does all this leave us as regards an overall evaluation? We still feel that the Cizek is one of the very best box speakers any where near this price. Not because it's without faults but rather because none of the others we're aware of have acceptable full-range performance. The Rogers, for example, sounds incomparably better but can't handle any power and lacks almost two octaves of the Cizek's bass range. It also costs more, at least in the U.S.A. The DCM Time Window is also significantly better but can hardly be called a box speaker and is priced very much higher. And so on.

Maybe we shouldn't have reviewed the Cizek in such fast company. Still, we feel that the bass alone is worth the price of admission and that the basic design is not only intelligent but also capable of evolutionary improvement.

Cizek Model #2

Cizek Audio Systems, Inc., 15 Stevens Street, Andover, MA 01810. Model #2 acoustic-suspension loudspeaker, 8268 the pair. Five-year warranty; manufacturer pays all freight. Tested #1130 and #1139, on loan from manufacturer.

Speaking of evolution, Cizek has already come up with a good example: their Model #2 is a considerably smaller and less expensive version of the Model #1 while still offering the same basic sound quality.

The main difference is the substitution of an 8-inch woofer for the 10-incher; in combination with the smaller enclosure volume this still keeps the system resonance at the same frequency, namely 39 Hz (or 38 Hz, according to Cizek). In fact, the Model #1 and Model #2 have virtually identical bass response profiles; the two-position Q adjustment is also retained on Model #2, except that the obtainable values are 0.57 and 0.9, certainly not a major difference. The tweeter and crossover network appear to be the same, but only a high-frequency level control is provided, without the contour control.

Both models sound extremely similar. On organ music and other complex program material that's rich in bass energy, the Model #2 poops out much more readily; it just can't handle power the same way and begins to distort. On the other hand, it has slightly less zing and glare at the higher frequencies-and there's a reason. The peculiar multiplication of tone-burst cycles occurs at around 2 kHz on Model #2, which is less irritating than 3 kHz ringing. Furthermore, the amplitude of ringing is quite a bit smaller. The 16 kHz peak is still there, though, and so is the differentiation of pulses everywhere except way off axis toward the woofer side. It's basically the same speaker.

And that makes it pretty good value in our book. Purists, on the other hand, need not apply.

Editor's Note: The latest Dayton Wright full range electrostatic speaker, announced for this survey on the back page of our last issue, should have followed here in alphabetical sequence.

Unfortunately, it wasn't sent to us in time for testing, despite a firm commitment from the company. We now have another firm commitment from them to make the speaker available in time for Part II of the survey in our next issue. If they don't come through, we'll try to obtain a pair anyway, one way or another.

Meanwhile, on the basis of admittedly superficial exposure to the Dayton Wright in four different places at various times, we're inclined to doubt that it would challenge the Beveridge as our top choice. But we're perfectly ready and willing to change our mind should our tests turn out to be contraindicative.

DCM 'Time Window'

DCM Corporation, 725 S. Division, Ann Arbor, MI 48104. 'Time Window' floor-standing loudspeaker, $660 the pair. Tested #906 and #932, on loan from manufacturer.

Here's the one that will recoup you for about 20 years' subscription to The Audio Critic: a speaker system that has no equal even at twice the price (at least none known to us). In this Part I of our survey, we place only the following speakers ahead of the DCM Time Window: the Beveridge Model 2SW, the Rogers LS3/5A (but only from, say, 120 Hz on up at moderate levels-or else with expensive sub woofers), and the Koss Model One/A (but only every other day because in many ways we prefer the Time Window). That's all. The Dahlquist DQ-10, for example, which was our reference speaker before we started the survey, ranks be low the Time Window in our current hierarchy.

There's some consistency to the above: every speaker that, to our ears, equals or surpasses the Time Window has outstanding response in the time domain. The Beveridge reproduces pulses most accurately; the Rogers is next; the Time Window and the Koss are close behind the Rogers and about equally good, but with quite different deviations from perfection. How about that? You'd swear that we measured impulse response first and then picked our favorites, but you have our word of honor that our preferences were firmly established prior to any laboratory tests, strictly on the basis of listening. We have a feeling that, in the not too distant future, speaker designers will be investigating impulse response as routinely as frequency response.

The DCM people are among the most fervent advocates of pulse testing; the very name of their speaker, Time Window, shows where their head is at. Their literature makes the most convincing case we've seen for the over whelming importance of the time domain; if you discount its quite mild and inoffensive commercialism, you can learn more from it than from just about any other popular explanation of speaker design. It shows comparative pulse tests on fifteen speakers, including the Time Window, and even if they manage to make their own product look a little better than our test results indicate, the brochure is an eye-opener.

The physical configuration of the Time Window is quite unconventional. Imagine a waist-high cylinder standing on the floor, except that only the back of it is cylindrical; the part facing you, the listener, is a triangular prism, edge foremost. On the two angled faces of the prism are the drivers, symmetrically placed: two 6-inch Philips woofers and two Philips dome tweeters below them (that's right, below the woofers, not above them). Near the floor are two ducted ports, also symmetrically placed. The cabinet is surprisingly light; you can carry it from one end of the room to the other without requiring abdominal surgery afterwards, yet it's quite rigid. Weird but effective design.

The proof of design is of course in the listening, and what a nice-sounding speaker it is.

The first thing that strikes you is truly convincing spaciousness and depth; when a singing chorus walks in from offstage in an opera re cording, you can almost tick off the yardage as they approach. Only the Beveridge is in the same league in this respect; the Time Window beats the Koss and all the others. Transparency and delineation of inner detail are of a high order, but not quite up there with the Beveridge or even the Rogers; the Koss, too, sounds more open and delicately etched in the upper octaves but not throughout its range. Left /right imaging isn't quite as spectacular as with, say, the Pyramid 'Metronome', but we find it musically satisfactory. In fact, the entire speaker sounds unfailingly musical at all times and remains listenable regardless of your length of exposure to it.

That doesn't mean, however, a complete absence of colorations or other audible anomalies. Far from it. The bass is a little funny for one thing; more about that in a moment.

The top end could be a little more extended and silky-smooth; that last touch of HF quality seems to be unattainable with the Philips dome, even though it's better than most and handles power very well. There's some raggedness in frequency response throughout the audio range, which may be the reason why the speaker is particularly sensitive to room placement;

you've got to make sure that the room reflections zig where the speaker zags. It isn't any thing serious (minor frequency-response problems seldom are), except in the bass and the lower midrange, where things get a bit thick and muddled from time to time. Carefully tuning the distance of the speakers from the back wall as well as the side walls will clear up the problem; raising the speakers a foot or so off the floor also helps. When everything is trimmed in, the lows are quite clean and tight, but don't expect stupendous pressure bass on organ pedals, bass drum or bull fiddle. If we were pressed to designate a nominal corner frequency for the Time Window's rather bumpy composite bass roll-off, we'd place it at 50 Hz. Maybe even a little higher. After all, it's not a large box.

We say 'composite' because the bass response of a vented box is the sum of the out puts produced by the woofer and the vent.

Ideally, these should be complementary, arithmetically adding up to flat response. In the case of the DCM Time Window they aren't.

Each woofer exhibits the classic vented-box response you'd expect, but the vents by them selves have perfectly flat output down to about 18 Hz instead of filling in for the woofers with a humped response where the woofers drop out. Somebody obviously thought that this was a very good thing, but of course it doesn't add up to flat bass response out of the total system. The DCM design philosophy, as we understand it, is that in order to generate flat response within four walls, a speaker should never be designed for flat response anechoic-ally (i.e., with near-field measurement). We disagree with this philosophy quite vigorously and feel that it may have a great deal to do with the temperamental behavior of the Time Window as regards room placement. We don't want to make too much of the whole thing, on the other hand, since the speaker is still a lot better with this flaw than others are without it.

The reason why it's better-why it sounds better-is that (a) it reproduces pulses very accurately, with only minor glitches, and (b) it exhibits little or no ringing on tone bursts. In other words, it doesn't smear the signal. (One of those minor glitches worth noting is a little negative blip out of the tweeter just before the leading edge of a positive-going 0.1 m-sec impulse. We mention it only because it's very neatly fudged in the DCM literature; visible but cleverly buried. You can also tell from their pictures that the various electrostatics don't have this problem, thus confirming each of our own findings. Naughty, naughty.) To sum up, the DCM Time Window is one of our happiest discoveries since we've started testing audio equipment, and we recommend it wholeheartedly to any music lover who isn't planning to spend thousands of dollars on his speaker/amplifier setup. It makes sense, how ever, to get a very good amplifier to drive the Time Window; not necessarily a very high powered one, since the speaker is quite efficient, but a really clean one because superior time domain resolution will show up the difference.

Our congratulations to Steve Eberbach, the young engineering partner of DCM, for what may very well turn out to be a classic.

Fundamental Research Fundamental Research, 1304 Success Street, Pittsburgh, PA 15212. 'The Low Frequencies' subwoofer, 3450. Tested samples on loan from manufacturer.

This is the best subwoofer we've tested so far, which by itself isn't a particularly great distinction, since the Janis W-1 and W-2 are badly flawed designs and the Dahlquist DQ-1W doesn't go low enough. The Fundamental Re search is a good unit, however, in any company;

that's why we didn't hesitate to include it in this survey. And it happens to mate very nicely with the Rogers LS3/5A (more about that in the re view of the latter below).

The Fundamental Research is a waist-high floor-standing subwoofer incorporating a 12 inch driver in a sealed box with an approximate internal volume of 3 to 3% feet. It's perfectly classic and straightforward in design except for a weird little slotted board that's nailed right on the front baffle across the face of the driver.

This doesn't provide any low-frequency loading but is there ostensibly to dissipate the higher frequencies-a very crude sort of termination.

It's responsible for the one serious flaw in the response of the woofer: a peak at around 450 Hz that rises 7 dB above the flat portion of the curve, followed by considerable raggedness (huge suckout at ca. 1 kHz, double hump back up to the 0 dB line at 1.4 and 2 kHz, ragged roll-off thereafter). Ideally, a woofer should roll off smoothly above a given frequency (450 Hz would be perfectly all right), so that the crossover network can do its thing without

running into unpleasant surprises. (Remember the 14 dB peak of the Janis W-1 at 460 Hz?) We must admit, however, that we could detect no audible anomalies above the crossover point when biamping the Fundamental Research either through the Rappaport PBC-1 passive crossover or the Dahlquist DQ-LP1 active/ passive crossover, with the nominal crossover frequency in the 100 to 120 Hz region. It appears that the goof has to be of Janis proportions before it can be heard.

That said, we give this subwoofer a high mark for its bass response profile. Down to the roll-off point it's dead flat; the -3 dB frequency is 38 Hz; from there on the response drops to -5 dB at 30 Hz, -8 dB at 20 Hz, and -14 dB at 10 Hz. This is a slightly overdamped profile (the Q appears to be a little lower than 0.707), so that there's probably room for further improvement (i.e., a lower corner frequency without the penalty of overhang) with some fine-tuning of the parameters. If you've read our subwoofer article in the second issue, and if we now tell you that the Fundamental Research handles power quite well, you need no further information to know exactly what it sounds like. The bass is deep, solid, well controlled-it couldn't be anything else.

Mother Nature practices no deceptions. (Or, as one of our consultants is fond of saying, the spectrum analyzer is too dumb to lie.) Overall, we prefer the Fundamental Research to the Dahlquist DQ-1W, even though it's much less efficient and therefore requires a bigger amplifier to drive it. But it holds up much better in the bottom octave and at the same time it's just as tightly controlled. Not quite as well built, though; it has a few cabinet buzzes, seldom excited even by loud music, that we'd rather do without. A little more wood or a little more glue seems to be called for.

We've left the best for the last. This sub woofer was designed purely by ear by a dedicated young man named Mike Zelenak. The chances that it would come out this success fully were slim, to say the least, but there it is --you can't argue with a historical fact. The moral is that, if your ears (and taste) are as good as Mike's, it's still possible in audio today to fly by the seat of your pants. Otherwise, especially if you're into woofers, stick with good old Thiele and Small. We will. But then we also believe in keeping road maps in our glove compartment.

Infinity QLS Infinity Systems, Inc., 7930 Deering Avenue, Canoga Park, CA 91304. Quantum Line Source speaker system, $2500 the pair. Five-year warranty, customer pays all freight. Tested #7001070 (left) and #7001071 (right), on loan from manufacturer.

In our first issue, we published a rather negative preview of the Infinity Quantum Line Source, based on superficial exposure to it in a dealer's showroom. After thorough testing un der our own roof, we have somewhat more respect for this speaker, but by no stretch of the imagination and with no amount of good will could we call it State Of The Art, Reference Standard, or any other such name dear to the hearts of audiophiles.

To put things into perspective, if our first exposure to the QLS, as inserted into our reference system, had been with a blindfold on, we would have said, "That's a fairly decent sounding speaker system; what is it?' On being told that it's Infinity's current flagship, successor to the fabled Servo-Statik and $1250 per side, we would have said, "You must be kid ding!" When properly set up and balanced out (which takes some doing), the QLS reproduces the full audible range of frequencies with a certain degree of authority and musical felicity.

But it never reveals sufficient inner detail (the pleasure derived is more in the area of maple syrup than of crystals); makes the stereo image shift and drift depending on frequency; woofs up and occasionally even burps when excited by really fierce bass transients; in short, it just doesn't sound like one of the more accurate systems money can buy.

The reason for some of these shortcomings is obvious as soon as you remove the grille from the front panel of this 5%-foot high, 150 pound monument. Have you ever seen or heard of a speaker with 17 (count them) drivers that was an accurate reproducer? We haven't. It's impossible to synchronize that many separate sources of radiation to produce a single, un confused image. Infinity seems to believe that by stringing out eight tweeters and six midrange domes in two parallel vertical lines they've created a Line Source (capital initials theirs).

Anyone who has read our Beveridge review above has some idea of what a line source really is and does. Where's the cylindrical wave front that's invariant with frequency? Where's the time coherence? We found that the QLS re produces pulses quite poorly, lacking even an acceptable "sweet spot," which is totally contradictory to the definition of a true line source.

What's more, the responsibility for balancing out those 17 drivers is left entirely to the user. There are three continuously variable level controls in the back (midbass, upper mid range, tweeters) with a rather wide latitude of adjustment; in addition, the tweeters can be brought in at three different crossover points by means of a plug-and-socket arrangement. Assuming conservatively that the continuously variable controls have only four audibly distinguishable positions each (say, all the way down, all the way up, and two in between) the total number of different-sounding response profiles possible is 192. The speaker designer is saying, in effect, "I don't know how the hell it should sound. You tell me, baby." There's a point where flexibility becomes anarchy. We tried our very best, but we can't guarantee that further knob-twiddling wouldn't have resulted in slightly better sound in our room.

Even with the help of a spectrum analyzer and a calibrated microphone, the overall frequency balance (i.e., far-field amplitude response) of the QLS is extremely difficult to determine on account of those 17 drivers strung out all over the place. All we were able to do was to analyze the tweeters, the midrange domes, the "midbass coupler" and the woofer separately.

We must admit that we were very impressed by Infinity's samarium-cobalt Electro magnetic Induction Tweeter (EMIT), a force over-area transducer claimed to possess all the advantages of electrostatics without their draw backs. We found no evidence to contradict that claim; the EMIT appears to be extremely smooth and flat out to 23 kHz, with wide horizontal dispersion. We haven't seen or heard a better tweeter, and that includes the rather similarly configured electrostatic tweeter in the Koss Model One/A. In fact, we got the most musical sound out of the QLS by selecting the lowest available tweeter crossover point and assigning the widest possible range of frequencies to the EMIT. That way the highs remained silky and sweet no matter what we fed into the speaker; furthermore, the EMIT reproduced pulses accurately as long as it was tested all by itself, with the adjacent midrange source block ed out. Now all that Infinity needs to do is develop an EMIT that goes all the way down to 20 Hz! The midrange dome used in the QLS is, on the other hand, a real turkey; it has, among other things, a huge peak at 2.2 kHz, which is probably one reason why the QLS doesn't have a more transparent upper midrange. As an interesting sidelight, this very same Peerless driver (Peerless is owned by the same parent company as Infinity) is available to all comers in the industry without the little pinholes in the dome that Infinity insists on having in the QLS version. To us the latter looked from the beginning like a cute little Helmholtz resonator-and sure enough, it resonates. We've heard it rumored that the Peerless people themselves don't quite under stand why Infinity wants holes in the dome.

Peerless, incidentally, also supplies the single 5-inch midbass (200 to 600 Hz) driver for the QLS, and it just so happens that this is the other slow member of the family (the wages of incest?), ringing like crazy at 200 to 250 Hz and contributing a retarded me-too response to 0.1 msec pulses, which are pretty far outside its as signed range. Talk about time smear . . .

Which brings us to the woofer-the same Watkins woofer we referred to in our preview as well as in our CES report. This isn't a unit to be dismissed quite as peremptorily as we originally did, even though the uncomplimentary remarks about it that we attributed to a distinguished American professor and electro acoustician are confirmed to have been uttered substantially as quoted. Another heavyweight academician, the polymath Dr. Richard H. Small of the University of Sydney, Australia, takes a somewhat kinder view of the woofer, as revealed in a letter to William H. Watkins, the inventor, a copy of which was made avail able to us by Infinity. More about that in a moment. Our own measurements indicate dead-flat response down to the lowest reaches of the audio range; the -3 dB point is 22 Hz! No other woofer known to us comes even close to this kind of penetration of the bottom octave, except the Janis W-1; and, as in the case of the Janis, we must point out that the audible performance doesn't live up to the promise of the response curve. The low bass is quite beautiful, detailed and controlled as long as the energy level remains moderate; however, when the woofer is called upon to move a lot of air (as in the case of one of our master tapes on which two double basses are plucked fortissimo), some rather rude eructations ensue, even with the best and most powerful amplifiers available. This doesn't surprise us, since a 12-inch woofer, Watkins or no, just doesn't have the piston area and linear excursion capability to find happiness in a 7 or 8-cubic-foot sealed box as a 22-Hz system-not at high SPL, anyway. Either the box should be smaller and the bass roll-off higher (cf. the Fundamental Research above), or the driver should be at least a 15-incher. As is, the system is really reaching when it digs into that bottom octave.

Even Dr. Small, of whose letter Infinity seems to be inordinately proud, points out that the Watkins woofer has no large-signal ad vantage over a conventional design. When it comes to moving air in a room, the cone alone must do all the work, and the limiting factors are cone size and linear throw. What Dr. Small likes about the Watkins invention (a two-voice coil configuration that eliminates the conventional hump in electrical impedance at the resonant frequency) is that under small-signal analysis it appears to provide an increase in out put of approximately 2 dB, all other things remaining equal-a theoretical free lunch that appeals to the academic mind. The penalty is that the two-voice-coil motor sucks current from the typical constant-voltage amplifier like there was no tomorrow, and the extra juice had better be available if you want to enjoy that 2 dB advantage. (Our Bryston 4B gave us no problem.) The amusing thing is that Dr. Small, a generous man who knows ten times more about speakers than Watkins, Infinity and The Audio Critic put together, peppers his mildly condescending praise of the design with quite a few sly professorial knuckle raps, which make his letter something less than a show-off piece encased in plastic for Dad's wallet. Certainly his exegesis of the woofer is at considerable variance with Infinity's simplistic technical brochure on the subject, which treats the low frequency impedance peak of a speaker system as some kind of foul disease for which a cure has finally been found. We have personally heard Neville Thiele, Dr. Small's original mentor, call that point of view "rubbish." An impedance peak and an amplitude response peak are two very different things. So much for the Infinity/Watkins/Small issue.

A word about the biamping option on the back plate of the QLS. If you connect separate power amplifiers to the woofer and to the rest of the system, without using an electronic or passive crossover ahead of the amplifiers, you gain very little in performance. Each amplifier channel will still see the full frequency range of the signal at its input and output, except that part of the output signal will be burned up in a network instead of being converted into sound.

For reduced IM distortion and increased head room, you need the additional high-impedance crossover. Of course, the fact that the separate terminals are there is a convenience. In any event, biamped or not, the QLS needs a lot of watts to drive it, mainly on account of that power-hungry woofer.

To sum up, then, the Infinity QLS gives you a big, juicy, quite pleasant and listenable sound, smooth and extended on top but far from transparent or analytical overall, with time smear quite evident and no stable image.

All pretty much confirmed by measurement.

Among the speakers we'd rather listen to are not only costlier ones like the Beveridge or competitors like the Koss, but also consider ably lower-priced ones like the Snell, Dahlquist, DCM Time Window and Rogers LS3/5A.

Koss Model One/A

Koss Corporation, 4129 North Port Washington Avenue, Milwaukee, WI 53212. Model One/A full-range electro static speaker, $2100 the pair. Tested early production samples, on loan from manufacturer.

Until the arrival of our Beveridge Model 2SW, the revised Koss electrostatic was more or less our reference speaker in the course of this survey. We say more or less, as we were never completely sold on it; however, in com parison with all the other speakers (except possibly the Rogers, which is a very special case), it just sounded more open on top, clearer, more analytical, less smeared. In other words, like a good full-range electrostatic. At the same time it had a disturbingly closed-down, thick, gag-in-the-mouth quality somewhere between the bass and the lower midrange-we couldn't decide exactly where. We experimented with room placement, biamping, taking off the back, you name it. Our lab tests finally revealed what we now believe to be the cause, but it wasn't something that could be helped without major surgery. We'll come back to that in a moment.

To understand the Koss Model One/A, you must realize that it represents just the opposite approach to full-range electrostatic de sign from the Acoustat, the Beveridge or even the Dayton Wright. It's much closer to the Quad, except for its four-foot height. The audio range is divided into four passbands and specialized electrostatic elements are designed for each, to be crossed over at 6 dB per octave.

There are four large bass panels, one mid range panel (about the same size), one narrow treble panel, and an even narrower tweeter panel. They're driven via transformers by any amplifier of your choice. The main difference between the original Model One and the One/A is that in the latter the treble and tweeter panels are placed next to the midrange panel (at the inward edge of each speaker, to form a mirror-imaged L/R pair), instead of partially covering up the midrange panel with disastrous effect on the sound. Either way, there's no serious attempt to time-align the panels; basically they're just fastened into the frame to fill out the available space. We suspect that this is one of the fundamental design limitations of the Koss, since the superior time-domain characteristics of each individual panel don't add up to coherent time response by the total system, except over a very narrow angle, way off toward the outside edge. The speakers would have to be angled inward to take ad vantage of this, not just a little but to an extent that wouldn't be tolerated in most homes. Too bad; the pulses are truly excellent from that angle but deteriorate badly as the head-on position is approached. We aren't suggesting that this isn't preferable to time smear that -originates in the transducers themselves;

the One/A has audibly better inner detail than nearly all other speakers, but the design concept isn't taken all the way to the limits of its potential.

In the frequency domain the speaker is also very impressive; it's possible to find a "sweet spot" where the response from 1 to 17 kHz is simply dead flat. The treble and tweeter panels appear to be truly superior transducers, except for a 6 dB peak at 15 kHz when the tweeter is in the head-on (axial) position. A number of listeners complained about an irritating quality on the top end when the speakers were played broadside; this may very well have been the cause-and one more reason to angle the speakers inward if space permits.

On the bottom we found the useful response of the One/A to extend smoothly to approximately 40 Hz, rather than 32 Hz as claimed by Koss, but then a dipole is notoriously hard to measure at the lower frequencies. (As a matter of fact, there exists no proven mathematical model for low-frequency propagation by a large, free-standing, planar transducer.) The subjective quality of the bass is by and large excellent: deep, firm and authoritative on program material that contains genuine LF information. Except . ..

That peculiar thickness or muffled quality kept bothering us, without much of a clue as to where to look for it. We finally nailed it-we think. At exactly 50 Hz, the entire frame of the speaker system takes off, creating a completely uncontrolled resonance and inducing the panels to resonate as well. The resonant band is quite narrow (high Q), so that it isn't obviously ex cited by typical program material, making it somewhat elusive without careful sweep testing.

Once found, the amplitude of the resonance turns out to be somethin' else: 8 dB for the integrated response of the four bass panels, 20 dB at the surface of the individual panels.

One bass panel we measured also resonated at 57 to 59 Hz; this may have been a problem within the panel itself, but the overall 50 Hz resonance appeared to be definitely a frame problem. With two strong men squeezing the frame together from both sides, the center frequency of the resonance was shifted by a few Hz without curing the problem. In our opinion, it would take a complete redesign, with much heavier bracing and generally sturdier construction, to eliminate the resonance. And we're pretty well convinced that the coloration we heard was caused by transient effects at the 100 and 150 Hz harmonics of the 50 Hz fundamental. Without this defect we'd be inclined to rate the Koss Model One/A just a hair be low the Beveridge Model 2SW, since it handles power a lot more comfortably than the latter even if it doesn't quite have the same super transparency. The way it was, many listeners preferred not only the Beveridge but even the DCM, which sounded somehow more spacious, open and natural in the all-important lower midrange.

The One/A also has separate terminals in the back, just like the Infinity QLS, to allow biamping the bass panels and the rest of the system separately if you so desire. The difference is that a convenient toggle switch is provided for instant change from the single-amp to the biamped mode. That made it very easy for us to determine that the difference (with one Bryston 4B vs. two Bryston 4B's) was barely perceptible without an electronic crossover.

The biamped mode was declared very slightly more open and transparent by all listeners after considerable agonizing. All our other re marks on the same subject in the Infinity QLS review above apply equally here.

We recommend the Koss Model One/A, then, to all those who can't live without the clarity, inner detail and generally un-smeared sound of a full-range electrostatic, but at the same time won't spend the extra kilobucks to go all the way to the Beveridge or to something even more elaborate like the HQD System (Mark Levinson's creation for oil sheiks). On the other hand, for a lot less money, a pair of Rogers LS3/5A's with or without subwoofers, or a pair of DCM Time Windows, will provide comparable (in some ways even superior) musical satisfaction.

Ohm F Ohm Acoustics Corp., 241 Taaffe Place, Brooklyn, NY 11205. Ohm F "coherent-sound" speaker system, 31200 the pair. Five-year warranty, excluding cabinet; customer pays all freight. Tested #32053 (cabinet #19297) and #32068 (cabinet #19296), selected at random in sealed cartons from manufacturer's stock.

Editor's Note: To prevent rumors, anonymous letters, crank calls, slanderous know-it-all comments, and other manifestations of ugliness and paranoia endemic to the high-end audio scene, the following disclosure must be made.

The Editor/Publisher was one of the original founders of Ohm Acoustics in January 1971 and for several years thereafter its largest single stockholder. His involvement in the company was strictly as a Director and nonresident consultant, without any day-to-day participation in management or engineering. In March 1974 he sold all of his stock back to the company and has currently no closer ties to Ohm Acoustics (or any other audio manufacturer or retailer) than to the Exxon Company or the government of Liechtenstein. He does confess, however, to continued partiality to the brilliant, if incomplete, loudspeaker theory of the late Lincoln Walsh.

The Ohm F is the great paradox of this speaker survey. Its ability to reproduce single pulses of different widths (in our tests between 1 msec and 0.1 msec) could conceivably be judged second to none, perhaps not even to the Beveridge. Its frequency response is also quite acceptable, except for a few anomalies noted below. Yet its sound is more highly colored ('"'canned' may be the better word) and less musical than that of any other speaker in this admittedly formidable group. Whenever we switched to it in the middle of a listening test, the instant reaction of those present was,

"Huh? What on earth is that?" The midrange, especially, sounded like a tin can next to the really accurate speakers.

Why? We think we know exactly why. The Ohm F rings. It rings more than other speakers. It rings like a telephone. It isn't even possible to localize the ringing at specific frequencies; our test sample rang virtually every where. At 11 kHz, where there was also a 6 dB peak in the amplitude response, our sample did worse things to a tone burst than we had ever seen in our laboratory. But then 11 kHz isn't in the most sensitive range of the ear. Maybe it was the 900 Hz ringing that was the real culprit. Maybe some other frequency. There were just too many bad spots to choose from. On top of it, the frame of the Walsh driver resonated at 118 Hz.

Interestingly enough, it seems that another investigator has come independently to the same conclusion. Long after our own tests, we discovered that the DCM 'Time Window' brochure uses the Ohm F as the classic example of ringing. Not by name, of course; but the oscilloscope photograph they use as their ex ample is identical to the one labeled Ohm F in their fifteen-speaker comparative pulse test further below on the same page. The photo- graph also shows that the Ohm F reproduces the initial pulse perfectly, exactly as we ob served, but then just keeps on producing an output without an input.

We don't want to create the impression that this defect of the Ohm F is a vulgar design error, such as we occasionally pillory in this publication. It's damn hard to make a Walsh driver that doesn't ring. The basic requirements of the cone present a conflict. Since the Walsh speaker has been around for more than five years, here's just a quick reminder of how it works: A single cone, mounted convex side out and apex up, acts as a transmission line (in the antenna sense, not the bastardized acoustical labyrinth sense). Sound travels down the side of the cone faster than in air, and through proper choice of geometry the horizontal component of the traveling wave synthesizes a coherent cylindrical wave front in the air, starting at the cone surround. That's the theory, anyway; its divergences from real life belong in another discussion. In any event, the cone material must be extremely stiff (i.e., have a high Young's modulus) in order to have the proper sound propagation velocity. That means a high Q. At the same time, the cone material mustn't ring; it must be well damped.

That means a low Q. Thus the ideal cone material for the Walsh driver would combine the most desirable characteristics of beryllium and mucus. No wonder that Ohm hasn't found it yet. They try to get away with a mundane combination of titanium, aluminum and paper in tandem, which doesn't quite make it. As we've said in the past, the Walsh invention would have deserved a multimillion-dollar R and D program to solve these and other immensely difficult design problems it poses. In which case it might have become the world's simplest and most accurate speaker. It's far from that right now.

A less serious objection we have to the Ohm F is that somewhere around 36 or 37 Hz, its response in the sealed enclosure has a 3.5 dB hump, meaning a system Q of roughly 1.4.

That's too loose and boomy in our opinion;

it represents the rock-pop taste rather than the devotion to accuracy one would expect of the knights of Walsh. (Unless, of course, the speaker is put plunk in the middle of the room, in which case the Q is approximately correct- but how many people will use it that way?) The -3 dB point is about 28 Hz, but that's 6.5 dB below the hump, so you don't get the subjective impression of a flat 28-Hz box. On the contrary, in nearly all room positions there remains some rather annoying midbass boom.

So, if you want one of the most brilliantly conceived speakers of all time-with a boomy bottom; a highly colored, ringing, metallic midrange and top end; plus superb time response if your ear could only separate it from all that mess-get the Ohm F. Incidentally, this is the "improved" version; some minor changes were made about a year and a half or two years ago. We remember the original version as having been better, but we didn't have all these other speakers to compare it to, and besides we don't claim an acoustic memory quite that retentive and precise. It's possible that our standards of excellence have changed more than the speaker.

Pyramid 'Metronome'

Pyramid Loudspeaker Corp., 71-07 Woodside Avenue, Woodside, NY 11377. Metronome Model 2 + 2W speaker system, $2525 the pair (in black formica, as tested, less in other finishes). Three-year warranty. Tested #169/269 and #170/270, on loan from manufacturer.

In our Summer CES issue we gave a quick preview of Dick Sequerra's latest all-out at tempt at SOTA and reported a favorable first impression of the Metronome speaker system.

Just how the CES units resembled, or differed from, our test samples we'll never know for sure; until very recently Dick was going through a modification-of-the-week phase, so the only fair way to structure this review is to report everything in chronological sequence.

When our test samples first arrived, we were assured that they were typical of the limit ed number of units already sold and then being produced. They turned out to be absolutely tops in our survey in two respects: dynamic range capability and left/right imaging (among the larger units, that is; the mini-systems are inherently excellent imagers). The Metronome was able to produce mind-blowing SPL peaks with suitable amplifier power behind it; it just couldn't be made to burp, grunt, spit, sizzle or show any other sign of distress no matter how hard we drove it. Clean as a whistle.

And left sounded like left even when we moved all the way to the right. Just like real life, you say? No, the Metronome didn't sound like real life at all. The top end wasn't really open;

the inner detail had no lifelike clarity and delicacy; everything was a little thick and smeared. We were both intrigued and disappointed.

In the laboratory, this particular version of the Metronome turned out to be almost perfectly flat in frequency response up to 15 kHz, at which point it cut off rather sharply. The only significant departure from a dead-flat profile was a 6 dB hump at 60 Hz, a little bit peculiar in view of the 37-Hz system resonance specified for the Model 2W subwoofer in the Pyramid literature. It seems that the response we measured, indicating a system Q of 2, was synthesized by a combination of the subwoofer, its roll-off network, and the lack of low frequency roll-off on the Model 2 top section that links up with the Model 2W. (The Model 2 is allowed to run wide open; it's a complete and listenable three-way system all by itself, with a somewhat overdamped 8-inch woofer.) Despite the 60-Hz hump, the subwoofer didn't sound boomy, just rolled off. No wonder: the -3 dB point was at 30 Hz, meaning that the response dropped 9 dB in the all-important octave from 60 Hz to 30 Hz. According to Dick Sequerra, this is a deliberate design feature to avoid subsonic garbage and acoustic feedback at the high SPL's the speaker is cap able of. We aren't sold on this philosophy, as you can probably imagine; nor do we believe that a 14-inch driver is correct for the internal volume of the subwoofer enclosure, which isn't much larger than that of the biggest bookshelf speakers. This is just the opposite of the Infinity QLS woofer approach: an oversize driver in an undersize enclosure. At least the oversize driver is never asked to move more air than it's able to, hence its apparent freedom from the usual eructations. But it can't give you the low lows, as it could with more volume behind it. All in all, though, this was one hell of a flat speaker system.

The speaker was also quite free from ringing throughout its range, but its problems be came apparent when-you guessed it-we test ed it with pulses. This particular version re produced a 1 msec pulse satisfactorily, got progressively worse as we narrowed the pulse to 0.4 msec, completely reversed the polarity of a 0.26 msec pulse, and was totally unable to reproduce a 0.1 msec pulse. A classic case of time smear. We were then told by Dick Sequerra that the 2-inch tweeter was undoubtedly wired into the system out of phase. We had him re verse the phase, after which the impulse response got worse. Dick then decided that the original wiring had been correct after all and reversed the phase back again. So much for version one.

Sometime later, we received a call from Dick informing us of a new 4-inch midrange driver that was supposed to improve the performance of the Metronome considerably. Our Model 2 top sections were taken away to be modified and returned to us with the new midrange installed. The results were unbelievably bad. The beautifully flat frequency response was gone; the midrange output was elevated and appeared to have extreme phase incoherency (180° out with respect to the rest of the system); impulse response was the worst yet-and the sound was definitely degraded, with unmistakable time smear. A few days later we got another phone call, apologizing for the horrible mistake; the midrange had been incorrectly modified and would we please re turn our top sections once more for the correct modification. End of version two.

Version three arrived shortly thereafter, with assurances that this was It-the final, irrevocable production model. All units already in the field would be modified free of charge to conform to this one, we were told. And sure enough, this new version sounded different and, to our ears, somewhat better. More open highs, inner detail more clearly etched, more air around the instruments and voices. All without compromising the unique dynamic range and imaging capabilities. We still preferred the Beveridge by a wide margin and the Rogers as well as the DCM by a narrower margin; they just sounded more real, with more of a see-through quality, even though the Metronome was objectively the "cleanest" in the sense that it produced no grit or crud, no matter how complex or dynamic the program material happened to be. At the same time, something was subtly wrong with this new version that we hadn't noticed with version one; the latter had been less transparent and real but somehow more consistent, stable and balanced in sound quality. We weren't sure what the difference could be attributed to.

We took the speaker to the lab and realized almost immediately what was going on.

For the first time, the Metronome was reason ably accurate in the time domain, with vastly improved results on the pulse tests, though still not in the Beveridge category. At the same time, the frequency response was completely shot to hell. So much so that it would be futile to try to describe its dips and bumps, since it was totally unrelated to any discernible zero axis. As a desperate example, there appeared to be a 10 dB peak at 6 kHz. With respect to 0 dB at what frequency? Don't ask.

What this proves is that good time response without good frequency response sounds more like real life than good frequency response without good time response. What it doesn't prove is that the Metronome speaker system is a finished product. We know Dick Sequerra well enough to predict that he won't be satisfied very long with the current production model. He is a perfectionist and wants basically what we want. He has accomplished a great deal already: reasonable size for a super system, good looks (we love that 45-inch tall metronome), fantastic construction (mostly 1%-inch walls!), stupendous dynamics, no gritty garbage out with no garbage in, precise stereo imaging, and now a fair degree of freedom from time smear. But that's not enough.

For two and a half grand, we expect more from the man who gave us the Sequerra Model 1 tuner than a 50 Hz to 15 kHz speaker system that doesn't sound quite as real as the little Rogers at one fifth the price.

Rogers LS3/5A

Reference Monitor International, Inc., Suite 309, 4901 Morena Boulevard, San Diego, CA 92117.

Rogers LS3/5A BBC Monitor Loudspeaker, $450 the pair (approx. $250 if purchased in Great Britain). Five-year warranty. Tested #S01777A/B and #S01807A/B, on loan from distributor.

The Rogers 1.S3/5A, a British import, was originally developed for the BBC as a very com pact monitor. Its outside dimensions are 12 by 72 by 6% inches. And it sounds better, at least from about 120 Hz on up at moderate levels, than any other speaker in this survey, except the Beveridge and the Koss-and we aren't so sure about the latter (see our Koss review above). For example, it's distinctly superior to the Infinity QLS.

How can a little shoebox do that? Well, how did a few Spitfires and Hurricanes beat the Luftwaffe? (And, rest assured, Hermann Goering was a harder nut to crack than Arnie Nudell.) Actually, old chap, it's bloody simple. You take the 4-inch KEF bass driver and the ¥%-inch KEF dome tweeter. They're jolly good, don't you know. You cross them over with a proper network made of decent parts (it's no place to skimp by using cheap electrolytics, etc.). You pay special attention to the time and phase characteristics. You make the sealed box as solid and rigid as Westminster Abbey. And then you cheat a little.

By cheating we mean the bass response profile. It's a bit of a humbug: would you believe +6 dB at roughly 130 Hz? Now that's a Q of approximately 2, and all the standard texts will tell you that the -3 dB point will then be around 75 Hz, which works out just about true to life. In other words, if the hill is high enough, its foothills won't drop out of sight, either. Is this any way to get bass response? We don't particularly like it; our strict upbringing makes us recoil from it-but how else are you going to extract a subjective impression of bass out of that itsy-bitsy box? Most people are fooled; time and again visitors told us that the bass sounded just great-until we showed them real bass. In all other ways, however, the Rogers has impressively flat frequency response, all the way up to the limits of hearing. By 20 kHz it's down somewhat, but then the English have always been kind to their dogs.

Our pulse tests revealed outstanding time domain behavior and also explained why the Rogers sounds best with the tweeter end of the box tilted backward or angled inward. The on-axis impulse response showed some anomalies, but 45° or so off axis the two drivers fell into a time-aligned configuration, rendering better pulses between 1 msec and 0.1 msec than any other speaker tested so far, except the Beveridge. In view of the latter's size and weight, the Rogers became our convenient laboratory reference standard for pulse shape and coherence. Even so, it did exhibit a slight negative blip from the tweeter at the leading edge of a positive-going 0.1 msec pulse; how ever, all other conventional speakers, even the DCM, were doing the same thing to a consider ably greater extent. Furthermore, the Rogers was remarkably free from ringing in comparison with the others. It's a delightful or scary little bugger, depending on whether you're an audio enthusiast or a speaker manufacturer.

A pair of LS3/5A's, separated by the right distance for your listening position and properly angled, will give you beautifully focused, crystal-clear, un-smeared, highly detailed sound reproduction, with all the spatial information (front/back, left/right) you could ask for. No reservations or qualifiers necessary. The Rogers will also resolve differences between amplifiers with ease and therefore deserves a very good one. The Norwegian Electrocompaniet power amplifier appears to be a particularly happy choice, since it can't possibly develop enough power across the speaker's 15-ohm impedance to hurt it (the Rogers is rated for 25 watts speech and music-not continuous power) and also happens to make the speaker sound superb.

(We blew out the woofer on one of our test samples with the equally great-sounding zillion watt Bryston 4B; don't you be that stupid.) All right, what about that bass? Some people have suggested using two pairs of LS3/5A's to extend it, but that's not the way Mother Nature works. All you accomplish is a 3 dB increase in level (and power handling) with basically the same response profile. And you screw up the time response with the multiple sources. Our suggestion: live with one pair, as is, or go to a biamped system with subwoofers. We found the Fundamental Re search to marry very well with the Rogers, as long as the nominal crossover frequency was kept in the 100 to 120 Hz region. Ideally, the crossover should be a little higher, say 200 or 250 Hz, to get completely away from the bumpy area of the Rogers; however, the Fundamental Research has some problems only an octave or so above that range (see our FR review above), so we had to stay lower. A properly crossed-over, biamped Rogers system with subwoofers promises to be one of the best money can buy; we plan to go more deeply into the subject and make more specific recommendations.

Meanwhile, we suggest you carry a pair of Rogers LS3/5A's around with you (in your car trunk, in an overnight bag, or whatever) when you go hi-fi visiting. After listening to your friend's (or dealer's) $2500 speaker sys tem, you connect your little shoeboxes and you've got an enemy for life.

Snell Acoustics Type A Snell Acoustics, 10 Prince Place, Newburyport, MA 01950. Type A loudspeaker system, 31370 the pair.

Five-year warranty; manufacturer pays return freight.

Tested samples on loan from manufacturer.

This is a unique speaker system, in more ways than one. To begin with, it's by far the best-looking large speaker we can remember:

an utterly simple, virtually seamless, upright brick of polished wood and stretched cloth, four feet high, two feet wide and just over one foot deep, designed to stand flat against the wall.

The cabinetwork and finish are of a quality hardly ever seen in the hi-fi business; further more, the unit separates into a lower section (housing the downward-facing 10-inch woofer) and an upper section (with 4-inch midrange and l-inch tweeter), so that two not very strong people, or even one in a pinch, can move it around without breathing hard. Just beautiful --though admittedly not the most important thing.

Well, how important is frequency response? All right, listen to this. Nearfield response of the individual drivers in their pass band is dead flat (and we mean plus or minus close-to-nothing, right on the zero line) from 38 Hz to 22 kHz. The bottom end rolls off to -3 dB at 28 Hz, -10 dB at 20 Hz, -20 dB at 12 Hz. On top, the response dips to -6 dB at 24 kHz. So, using the convention of -3 dB corner frequencies, the total passband can be called 28 Hz to 23 kHz-and no ripples. How about that? We've never seen anything like it. What’s more, all drivers radiate this output without significant peaks or dips over a very wide angle, so that with the crossover used the power response into the room is also astonishingly uniform. Distortion is low; power handling is very good; the woofer is particularly impressive-the driver and the enclosure seem to be exactly right for each other. (All you Q = 2 guys, please note.) So how come the Snell isn't our top choice in this survey? You already know the answer if you've read this far in our alphabetical sequence of reviews: the frequency domain isn't where it's at. Mind you, the Snell sounds excellent. If we hadn't been exposed to the Beveridge, the Koss, the Rogers and the DCM, we would have lived with this kind of solid, wide-open sound quite happily for a long time. It's just a wee bit on the bright side (probably on account of the uniform power response right up to the highest frequencies), but we don't really mind that.

Besides, the tweeter output can be reduced at the flick of a toggle switch. And the overall sound is truly clean and balanced. But some thing is lacking. It's that utterly alive, high ceilinged, see-through quality and unsmeared resolution of inner detail available only with more or less time-coherent systems. Which the Snell isn't.

In fact, the Snell can't reproduce single pulses at all. You feed in one pulse, and out come two discrete pulses-one from the mid range and one from the tweeter. The two simply aren't joinable, regardless of distance or angle.

This is an inherently noncoherent system, probably on account of the large acoustic separation between drivers. Too bad, since it's obviously a system designed with tender loving care.

In a way we're glad, perhaps at the expense of Snell Acoustics, that this speaker is such a pure embodiment of the frequency-response above-all approach. If that were indeed the correct design philosophy, there couldn't possibly exist a better-sounding speaker, and we'll soon hear from some highly vocal sources just how wrong we are and where we should go to have our ears examined. On the other hand, if we're right-if time response is the more important design consideration-then the Snell Acoustics Type A will have served as the classic control experiment. Either way, it's a far from negligible contribution to speaker design.

Recommendations Please remember that the 15 speakers re viewed here, plus the dozen or so other units we've commented on in the past, represent only a small fraction of the universe open to serious audio enthusiasts. We can't swallow that whole universe in one issue-or even in four or six.

If you're looking for infallible and all-encompassing guidance, you've come to the wrong place. On the other hand, we know the stuff we've tested pretty well and can offer the following circumscribed recommendations with some degree of confidence.

Best speaker system tested so far, regard less of price: Beveridge Model 2SW (with reservation about headroom-see review).

Close to the best at a much lower price: Rogers LS3/5A (but bass not quite adequate without expensive subwoofer-see review).

Best full-range system per dollar: DCM Time Window (supersedes Dahlquist DQ-10).

---------

[adapted from TAC, Vol.1, No.4: July/August /Sept 1977]

---------

Also see:

Box 392: Letters to the Editor

Various audio and high-fidelity magazines