At times, it takes a certain constellation of events to finally flick that upstairs light bulb. Good advertising campaigns are based on that notion. But even advertising peppered with apparent implausibilities can do the trick. What prompted today's lineup of synchronicity triggers? It was a combination of factors: A recent SoundStage! review of the new VR-1 monitor speakers and a subsequent visit to its maker's website; Jim Saxon's brilliant column on White Papers, a humorous while poignant exposé on audio's propensity for creative claims; and finally, my recent Auroville foray into how patents don't always mean what they seem to imply - namely official seal & sanction on the merits of an invention or new application.


What follows are personal road blocks encountered on the rocky highway to enlightenment while perusing VSA's website in the wake of the recent review. I'll first give you the reasons why certain descriptions or suggested meanings struck me as obfuscation or being in downright opposition to textbook Physics. Part-2, Speaker Physics!, will feature Albert Von Schweikert's hopeful removal of these road blocks by way of succinct explanations. This exercise in admission of ignorance, expected clarification and regardless learning might be of help to fellow 'philes who occasionally find themselves equally befuddled; by fancy semantics or colorful acronyms, two ingredients our High-End audio circus certainly plays to on a regular basis (and often to cheering crowds waving large dollar bills.)


My first stumbling block? A loudspeaker crossover dubbed Global Axis Integration Network or GAIN for short. It purportedly controls a speaker's dispersion pattern such as to inversely mimic a microphone's cardioid pickup shape. I quote from the website:


"Albert Von Schweikert's design goal for the last 20 years has been to recreate a simulation of a live musical event. Not satisfied with conventional technology that he felt had been exhausted, AVS's recent research indicated that more realistic sound quality could be obtained by replicating an acoustic inversion of the microphone's signal. After all, it is the microphone's signal which is recorded, not music itself. Since microphones impart their characteristics upon the music, recording with microphones may be seen as an encoding process, while playing back the signal through speakers may be seen as a decoding process. If you think about this, you will realize that the goal of the speaker should be to behave as a microphone in reverse."


Thinking about the implications as prompted, the potential for steering sound to produce specifically contoured radiation patterns with a mere crossover pointed at a cure for beaming; for correcting lobing perturbations; for ultimately even making toe-in of speakers and sophisticated diaphragm geometries unnecessary. If once could simply dial the necessary offset variables into the crossover, then equalize for any dispersion geometry necessary? Eureka. Consider the licensing fees alone.


Unfortunately, this recipe nearly sounded too good to be true. To visualize my perplexed efforts at understanding the Physics involved, let's use a simplified two-dimensional model of how waves propagate on water. For this specific consideration, the longitudinal pressure waves of sound through air do behave in analogous fashion to the transverse surface ripples of water. Though note that this comparison won't hold - um, water as a three-dimensional model.


Let's drop a stone into a calm lake. What's the dispersion pattern? Circular. What if someone now told you that he had found a magical new way to throw a stone into a lake such that the normally concentric ripples would appear in the shape of a star, heart or triangle? You'd scratch your head. Isn't this very similar though to claiming that a mysterious GAIN circuit can focus and control driver radiation in a particular shape?


We'll briefly back up for a bigger picture. Say we dropped two stones into the lake at once. We'd get interference pattern where the ripples clash. Certain ones would cancel out. Others would add. Natural barriers such as a boat, the shore line, pilings of a pier, the edges of a speaker baffle and cabinet likewise would introduce distortion patterns in the natural concentric waveforms. This visualizes how driver spacing and shape, baffle size and geometry interact to affect dispersion. But using a round rock to create a distinctive wave pattern other than circles in unobstructed water? That would require certain waves to travel faster or slower than others. For the 'heart-shaped' pattern of the cardioid microphone above, we'd need consecutively slower waves for the 0°-90° and 0°-270° sectors and true snail pacers for the 180° axis. Put differently, we'd need to overcome the constancy of speed whereby waves propagate on water or sound travels through the air.


Unfortunately for our promising GAIN concept, no matter what we manage to do to our audio signal while zipping through the filter network (separating different bands, compensating for gross nulls or peaks, shaping the rate of attenuation), once the electrical signal converts into mechanical motion (the driver diaphragms acting like pistons pushing and sucking air), the resultant pressure waves propagate at one unalterable, hence constant speed: Roughly 13200 inches per second, the speed of sound.


How does that translate? In our simplified two-dimensional model above, the X-crossing of the intersecting 0/180° and 90/270° axes becomes the drive unit, the 90/270° plane the speaker baffle. Unless we operated our conventional round loudspeaker drive unit into the range where wave lengths and driver diameter began to coincide and beam, we'd expect circular waves. Unlike the cardioid shape then, this means equidistant radiation from the central X-crossing, not a contoured mushroom pattern. In three-dimensional reality, it of course also means specific vertical interference patterns as introduced by multi-driver interactions, floor/ceiling boundaries, baffle/cabinet diffractions, turbulences and reflections.

Back to the subject at hand. Can our GAIN circuit produce the cardioid or any other arbitrary dispersion patterns which diverge from the specific one predicated by the above variables (most of which, incidentally, occur beyond the filter network)? Let's ask this differently to highlight my confusion even further: Can the pre-driver filter correct for influences that affect the signal post-driver, after-the-fact, once it has scurried past the crossover to face the room from its position inside the speaker baffle? I don't understand how. Hence I'm eagerly awaiting Von Schweikert's clarifications.


Elsewhere, we read about phase- and time coherence with designs sporting 4th-order networks, vertical baffles and, in certain cases, rear-firing ambient tweeters. Vance Dickason's speaker design definitions distinguish between the terms phase coherence or phase linearity and being in-phase. While the latter is possible with our 4th-order network, the former is not. The Mayan step-pyramid scheme of Vandersteen speakers; the sloped baffles of Meadowlark, Sonus Faber, Thiel; the adjustable cabinets of Ascendo and Green Mountain - all manufacturers intent on phase-coherence ... they all account for these two requirements: Physical alignment of the drivers' acoustic centers & a 1st-order network. Instead, if we use vertical baffles as the VR-2 below (without recess or protrusion driver-mounting as the now-defunct Dunlavy implemented correctly) and add steep attenuation rates with 360° phase rotation? We're in defiance of the twin mandates of said principle.


But wait? Can't our filter network be engineered to introduce the precise delays in time which our baffle doesn't account for in physical offset?


We remember that our 4th-order network introduces a 360-degree phase shift, i.e. the waveform, at that specific frequency, is delayed by exactly one cycle. That makes it in-phase but not in-time. Therein lies the crux of the matter. This delay in time is a function of frequency, not a fixed temporal value across the band. Say we determined that the physical tweeter on our vertical baffle needed correction by one inch to be aligned with the woofer. Wouldn't a time delay of 76 microseconds accomplish that (1 inch divided by 13200/sec = 76 microseconds)?


Unfortunately, the delay of our 4th-order network, within a customary tweeter bandwidth of 2,000 to 20,000Hz, creates from 0.5 inches to 6.6 inches of delay depending on frequency (for example: 1 second divided by 2000 x 132000"/sec = 6.6 inches). In short, the 76ms/1" offset we determined as necessary for proper timing cannot be successfully implemented across the tweeter's operational bandwidth. You know the old proverb about the broken clock? It shows the proper time twice a day. Ditto for our attempted time correction scheme - we can fix it for but one frequency while still wildly missing the target for all others. That makes claims for time coherence with 4th-order networks and vertical baffles confusing. Never mind that a rear-firing ambience tweeter seems to obliterate the entire concept from the start. There's a related issue. Cascaded 1st-order networks are said to combine or add up to 4th-order filters. I quote:


"There are two schools of thought regarding the correct type of crossover circuits that should be used to blend the drivers. The followers of 1st-order filters believe that a simple circuit will sound more transparent, and that the natural phase-coherent behavior of these circuits will somehow impart a purer sound quality than higher-order filters. On the other hand, the engineers employing steep filters believe that the high degree of overlap caused by 1st-order filters between the woofer and tweeter can lead to distortion, coloration, and off-axis response errors. The truth is: both of these positions are correct! The choice of 1st-order or 4th-order filters are both compromises, each with a list of benefits and disadvantages. For this reason, Von Schweikert spent fifteen years developing a proprietary filter topology that would be a best possible compromise between the two filter designs. In effect, the Global Axis Integration Network employs ladders of first order filters, ensuring purity and coherence. However, the use of several stages of first-order filters also enables our GAIN circuit to have the benefits of fourth order filters, with lowered distortion and greatly improved off-axis response behavior."


Doesn't this translate as implied benefit of retaining the steeper attenuation rates of the 24dB/octave filter while simultaneously maintaining the phase coherence of the 6dB network? Doesn't the former's phase rotation -- intrinsic to the rate of roll-off -- make phase linearity and its associated benefits an absolute impossibility? Add that cascading just one single capacitor (such as you'd find in a basic 1st-order tweeter circuit) by a factor of four causes these stringed caps to act like one compound capacitor of very small compound value - a simple 1st-order high-pass assembled of more parts than necessary. Ditto for cascading the inductor of a basic 1st-order woofer circuit. It makes for one big inductor, not a 4th-order network. It remains true, however, that a 4th-order filter is accurately described as two cascaded 2nd-order filters. Do you appreciate the quagmire music lovers and audiophiles can sink into when paying enough attention to perceive conceptual morass but not armed with enough engineering wits to pull back out? I'm certainly stuck. Help!


My simple questions for Albert Von Schweikert are thus:
  • How exactly does the GAIN circuit steer the drivers' radiation of VSA speakers to combine, across the full bandwidth, into the depicted cardioid pattern? And what, exactly, is the benefit of this particular pattern to begin with?
  • How can the textbook terms phase- and time coherence be applied to 4th-order designs with vertical baffles?
  • How do cascaded 1st-order filters make a 4th-order or quasi 4th-order network? How does the resultant hybrid circuit combine the "best of both worlds"?

Having seen a preliminary while less civilized copy of this feature, Von Schweikert called to volunteer a four-hour personal elucidation and dinner during the upcoming San Francisco show. Seeing how just then, he'd have more pressing matters to attend to than enlighten a dim-witted while pesky reporter -- and remembering that said reporter would be hustling the hotel floors for his upcoming show report -- I opted for a subsequent e-mail format instead. Rather than relying on writing out what and how I understood his explanations (or thought I understood them), Albert could pen his own reply, in his own precise and undiluted words. He already did confess to a Bob Carver-like propensity for creating colorful terms, to describe what he still insisted were "truly novel inventions requiring new terms to distinguish them from the old". But he also promised that there was rock-solid science and math behind these apparent contradictions. While claimed to be patent-pending since his closure of the Watertown facility, he offered explanations on the operation of his mysterious circuit in clear and plain English. He then also invited me to listen to the newest VSA speakers at HE2003. Perhaps he missed the boat on that count though?


You see, none of these three questions (nor what motivated them in the first place) is concerned with subjective audible performance - even tangentially. For the purposes of today's discussion, performance isn't the issue. At all! My issue is with what I personally perceive to be overly creative marketing propaganda. It comes across as a questionable way of promoting and distinguishing one's products, especially if they were as good as the VSA reviews and awards have it. This undermines even further the one badly shaking leg High-End audio defends as its marginal perch with many engineers in other walks of life, professional audio included. One thing our ailing industry doesn't need more of is real or fake snake oil - because just the perception of abracadabrant gobbledegook (despite things really being in order) will have people react and judge accordingly. So stay tuned as Von Schweikert sets the record straight in our next installment. Perhaps some other speaker designers or engineers will see fit to contribute relevant insights and help disentangle our confusing knot of terminology, filter topologies and what they can and cannot do?