Interview with Albert Von Schweikert, Part 2
Given the response to the initial interview with Albert von Schweikert in Issue 1, we decided to pose a few more questions to Albert. Fresh back from RMAF where his VR-33 and VR-35 designs were drawing many positive reactions, we invited other owners and interested parties to ask questions, and pushed Albert on his design process and the future of VSA. As usual, Albert didn’t hold back and provides here perhaps the most detailed interview ever on his history, practices, and future. The beauty of HiFi Zine is we are not constrained by page limits or layout so here’s the full version, in all its glory.
PD – The VR33 seem to be causing a splash and went over very well at RMAF this year. Are you planning a smaller speaker along these lines?
AVS – We are in the midst of designing a smaller model with an 8” Scanspeak woofer and 1” Scanspeak tweeter, to be named the VR-22. It will have twin crossovers selectable by a rear panel switch, so that it can be placed against the wall or placed in free space. The “Boundary” crossover will have a lighter, tighter bass that will sound great next to a wall, and the “Free Space” crossover will feature louder bass for placement several feet into the room. The “twin crossover” idea was suggested by several customers who wanted to purchase the VR-33 but were forced to put them in between two rooms, where no boundary loading existed. Although this is a very rare placement consideration, I decided to offer this as a standard feature on the new VR-22, which is slated to retail for $2,000/pair. It will also feature time alignment and a phase-coherent wave launch (using a mechanical step-back design as introduced by the Dahlquist DQ-10 in 1976).
Although our website does not show the UniField Model One bookshelf/monitor speaker yet, it has been on the market in Europe and Asia for over a year. This is a “no-holds-barred” design using a 2.75” triple-laminate cabinet wall design and a BBC concentric arrayed woofer/tweeter unit in a small transmission line. It is Phase Coherent and Time Aligned and retails for $4,000/pr. The October/November issue of the UK magazine HiFi+, gave it a rave review, stating it was one of the very best monitors that they had ever tested, easily beating monitors at twice its price. I won’t go into the review here, except to state that if you want to see how much progress we have made since our 2003 VR-1, you’ve got to read this review!
You will see the UniField Model One on our website within one or two months, we’re pretty slow with our website work since most of us work 12-16-hour days.
PD – Some suggested that VSA speakers seem to be continuously in a state of improvement e.g., Mark 2 versions quickly follow, new parts are introduced, and so on. Do speaker designs evolve that quickly due to improvements in parts or does this reflect something of the VSA approach to continuous tweaking?
AVS – While it is true that our company seems to come out with upgrades every two or three years, this is the norm, not the exception, with every audio company that we know of. Since there are rapid developments in transducer design and new insights into cabinet damping every few years, we necessarily need to take advantage of this technology unless we want to be left in the dust by our competitors. To take a specific example, our VR-4 SR model is only in the Mk3 stage, while a model from a well-known high-end company that debuted at roughly the same time is in its 9th incarnation. Interesting, yes?
More importantly, most other speaker companies do not offer affordable upgrades, forcing the customer to sell their present speakers at a loss and purchase the newer model. Although our website does not specifically list the fact that we offer upgrades, hundreds of customers have discovered with a simple phone call that our upgrades are affordable and offer a tremendous value. For instance, our original VR-4 built in the mid 1980’s can be brought up to today’s VR-4 SR Mk3 ($14,000/pr) for only $4,000/pr. How is that for value?
Perhaps people believe VSA requires many rounds of engineering before a product is considered to be finished but that is hardly the case. When you go back to the mid 1980’s when we started out and discover that our models have an average life span of about five years, which I believe is higher than the industry norm. However, this might not be obvious. Another well-known speaker company has a wonderful little speaker that, while it appears to be the same speaker that debuted in the mid 1980’s, has actually undergone 19 changes, without any notice or publicity. They of course don’t want to become known for making changes every few months or every year or so, since they want to give the impression that they are selling “finished” products. We could do the same thing, but then our customers would not know to take advantage of the upgrades. Interestingly enough, that other company does not offer upgrades – you have to trade your old ones in to get the new model, if you even knew that there was a change that you wanted to take advantage of!
PD – We received a lot of questions about how you design, and what amps form part of your test equipment. Beyond these specifics however, more broadly there is interest in the VSA development and evaluation process. How do you characterize what you do and what makes a VSA speaker?
[Interviewer note: What follows is a lengthy explanation that is worth reading fully.]
AVS – The (long-term) development of my speaker designs took place in two areas: measurements and listening tests, occurring simultaneously as we developed our test regimens. At Cal Tech, a group of us used the electronics and acoustics labs to build a set of speakers for our personal use. We didn’t really know how to design a good sounding speaker, so we borrowed “famous maker” speakers and took them to the lab to measure them for “inspiration” and to gain a knowledge base of what types of technology had the best response at any given frequency range. At the time, no single design had good sound or measurements in every area, so we had to determine what types of designs had the best bass, midrange, treble, and the best spatial replication. At the same time, we were learning how to use the test equipment and run scientific experiments to validate our listening tests. Initially, we spent most of our time measuring data, but soon realized that we didn’t know how the measurements correlated with good sound. I’m sure you have all heard speakers that are “accurate” but not very musical, and vice versa. There is a delicate balance between all of the electro-mechanical elements in a speaker “system” that influences the listener’s emotional response. We quickly discovered that no-one knew how to design a speaker that sounded like a live musical event.
The second part involved learning how to listen, so we tried to correlate what we were measuring with what we thought we were hearing. We discovered that some great measuring speakers didn’t sound “real” and vice versa so we spent a couple of years trying to tie together the measurements versus listening experience. That is when we met Dr. Richard C. Heyser, who worked across the street at JPL and who designed much of the electronics for the Apollo lunar flight. He become the Chief Speaker Tester at Audio Magazine and did the famous Quad electrostatic review, all the while working as a NASA scientist at JPL. He as also the inventor of a significant test apparatus called the Time Delay Spectrometer, a device that could enable acoustic measurements without an anechoic chamber and could also measure phase data separately from the pressure (frequency response) data. After Richard became my mentor, he showed me a manuscript he was working on which attempted to correlate measurements with actual listening tests. His work indicated that it might take approximately 2,000 different tests to “map” a sound field. Although this statement put me off for quite a few weeks, I didn’t let this interfere with my plans to develop my own testing regimen.
From 1976 to 1978, we used a Tandberg TD20A open reel tape deck and “master tapes” to listen to our speaker designs. We also had a Garrard turntable with Shure V-15 magnetic cartridge and played records. What truly annoyed me was that we had collected several different “reference” recordings of symphonic pieces, and each record label had used different halls, mics, and orchestras. Since all these recordings sounded different from one another, we realized we were chasing our tails trying to use (commercial) recordings to analyze our speaker designs, since we couldn’t really know “how” the recording was supposed to sound. Trying to use other brands of speakers as “references” for certain frequency ranges also proved to be futile, since we didn’t know which speakers were the most accurate to the signal – we could only determine which designs we liked the best, but some of our reasoning was based on purely emotional considerations.
The breakthough came after building a large PA system for a local club and realizing that we could use live musical sources to compare speaker designs, since it would would ensure reliability and repeatability. These live sources were small portable instruments, such as acoustic guitar, accordion, harmonica, trumpet, clarinet, cymbal, snare drum, banjo, and naturally, both male and female singers. I assembled a listening panel of a few friends who played music and were also audiophiles, and we divided the room into two sections, one where we placed the instrumentalist or vocalist, and on the other side of the acoustic barrier (which we built from large sheets of R19 fiberglass 3 feet thick), we placed our speaker. We used a variety of different brands and types of microphones, in order to avoid biasing the speaker’s sound on the mic response. As I recall, we had Sony, AKG, Neumann, and RCA mics of both condenser and ribbon design, some with adjustable cardioid and omni patterns. We used a Sony solid state mic preamp and a Crown DC-300 transistor power amp. In retrospect, I doubt that either the preamp or amp can be considered to be “state of the art” audiophile units, but they did offer low distortion and repeatable results. Also, we tended to worry about our choices of microphones, since we realized that the mics might not be any more accurate than a crude speaker, but that turned out not to be the case – many microphones available then (and today) are of “reference quality” and have almost no coloration or distortion when compared to any speaker system. However, in order to negate the mic’s signatures, we used “arrays” of mics that each complemented each other to neutralize and minimize the effects of using a single microphone. How we accomplished this was the time consuming process of switching mics, making sure that each mic was in the same place, so that the time and phase of the speaker-to-the-mic was repeatable.
We then designed our speaker in stages, starting with the midrange drivers and appropriate enclosures. We were using the Quad electrostatics as our reference but the compression and lack of dynamic range left us unsatisfied, even though they were then the best sounding speakers on the market. It was easy to acquire cone midrange drivers at no charge from almost every manufacturer by listing our student status at Cal Tech, so we were able to test more than 50 cone units over a period of 6 months. We tried every type, including Kevlar, recently introduced by Davis of France and B&W of England, carbon fiber, every imaginable type of treated paper cone, and a few prototypes we built ourselves using paper as a base, laminated with surface layers of aluminum foil, rubber cement, balsa wood strips, ceramic and/or resin topical layers, etc. It became clear that a laminate or composite of different materials sounded far better than any one material, so we began to seek out engineering partners at driver companies to help us refine our ideas. Audax of France was one of the first major companies that put time and money into helping us develop a highly refined midrange driver, since they believed in our theories and felt we were helping them as well. Using several different elements in the cone was helping to tame peaks and colorations due to cone flexure, since the laminations damped the peaks by use of the “differing Q” principle, i.e., when one material is going into resonance, the 2nd and 3rd lamination is going into resonance at a different frequencies, helping to cancel and smooth out the peaks. It took 3 years to develop the first midrange driver that we thought was acceptable as a “reference.”
We realized that the enclosure and crossover were equally important. A huge finding for us was realizing the inside of the enclosure had to be completely resonant-free or the resulting echo would come back through the cone, causing coloration. This discovery was hastened by my idea to put a mic inside the speaker enclosure and measure what was going on inside the box, which let to my discovery of Gradient Density Damping, laminated cabinet walls, and specific ratios of length, width, and height of the enclosure, using internal baffles to break up the standing waves.
Another discovery when playing with crossovers was that no crossover was better than any type of crossover, unless the drive unit had large peaks that required taming, so we chose to design nearly full-range speakers and cross them over far below the ear’s sensitivity range. Most speakers of the day used 500Hz and 3kHz cross points, but we discovered that running a woofer up to 500Hz led to enormous coloration in the lower vocal range, so we pushed the crossover frequencies as far down as we dared without introducing bass overload. After several years of refinement, we believed we had designed a midrange driver and enclosure that was almost as neutral as the Quad, but had far more dynamic range and power handling. The “live” factor was tactile with the cone unit, since it had far higher efficiency (sensitivity) and had more of a realistic “jump” factor that we highly prized on jazz instruments.
By comparison, the Strathearn ribbon speaker, Quad and Janzen electrostats, along with the KLH 9 electrostatics, sounded compressed and “dead.” We traced back the compression and “lack of life” to the extremely limited mechanical excursion potential of a stretched thin membrane, so we were happy that in one important respect, we had out-engineered the fabled electrostatic and ribbon designs.
The eventual midrange driver was stunning on vocals, and in a direct A/B/X comparison test, sounded the closest to a live feed, while also outperforming all other contenders by a wide margin. Many of the mid range drivers in commercial use had colorations that were severe, such as a “honk” or “quack” on some instruments, shrillness on upper registers of the female voice, or “boom” on the lower end of male vocals. A truly neutral midrange driver was the most important accomplishment we made in several years of research, since no matter how fantastic the bass and treble is, a speaker is “made or broken” by its midrange response.
Next up, we had also acquired almost 100 tweeters of various designs, including domes made from every imaginable material for the day (pre-1980), a Walsh 360-degree bending wave tweeter, RTR electrostatic tweeter panels, and a Decca London ribbon tweeter. We had also extensively auditioned the Hill Plasma flame tweeter, designed by Dr. Alan Hill, which used burning helium as a plasma source, which in turn was modulated by the amplifier signal. This tweeter sounded best of all, but alas, was extremely impractical, so it disappeared off the market quickly. After experimenting with different types of dome tweeters, we initially used aluminum domes damped with a light rubber coating for almost a decade, then switched to treated silk dome units made in Europe. As with the midrange tests, we listened to natural instruments, then played the live instrument through various tweeters we wanted to compare. At first, it seemed that the hard domes (metal, plastic, or ceramic) had the most “sparkle” and “air” but that turned out to be an illusion, based on certain emphasis on harmonic overtones that almost sounded “better” than live. In the long term, the hard domes “rang” at certain frequencies, creating long-term listening fatigue. While the rubberized silk domes seemed to be a bit smoother and warmer than an actual cymbal, they were perfect on voices, acoustic guitar, and violin. In certain cases, we made conscious trade-offs based on long term emotional response, rather than immediate impressions. In the past few years, we have tested diamond, beryllium, titanium, and other types of extremely stiff and hard materials, and although there were some very good aspects to these types of tweeters (better off-axis response, lack of diaphragm break-up, and extended high frequency response) these materials also came with a dramatic drawback – they all “rang” at certain frequencies and were very fatiguing to listen to.
Simultaneously, several European companies had re-engineered their soft dome designs, eliminating most of the prior limitations of soft domes. Naturally, all of you reading this journal of my experiments realize that all forms of engineering require trade-offs in order to make the device work. While it is true that hard and soft dome designs both have their imperfections, long term listen ability is perhaps one of the most important aspects of speaker engineering that will either “make” or “break” the success of the speaker system. Not one person on our audiophile listening panel can listen to hard domes for more than a few songs before listening fatigue sets in. This is an “age” related problem, since older audiophiles with significant hearing loss tend to like the additional hard “edge” that they can’t normally hear with smoother tweeters. Younger listeners and musicians tend to like the soft dome tweeters due to their lack of listening fatigue and their greater fidelity to a live instrument. If you have heard a female vocalist “hitting” the mic hard, you will instantly hear the overload and break up that occurs with virtually every hard dome on the market. Since I have very sensitive hearing, all types of distortion can grate on me after a few songs, and during the A/B/X comparison tests of all the tweeters versus a violin, guitar, or female vocal, the soft dome wins every time!
A major stumbling block was the integration between the midrange and tweeters’ dispersion patterns, so we adopted Time Alignment, invented by Edward Long for the JBL/Urei series of studio monitors. In essence, the waves from the tweeter should appear to emanate from the same physical place in space as the midrange waves, and if you can’t mount the tweeter inside the midrange, you’ve got to find another method to integrate the phase of the two pressure waves. This is done by using mechanical setback of the tweeter, or by the use of delay filters in the crossover that result in phase shift (used to delay the impulse response) to achieve flat frequency response over a wide vertical and horizontal window. The crossover filter used to integrate the two drivers can be as simple as a single capacitor on the tweeter to block the bass, or it can be so elaborate as to use up to 18 or 20 filter parts to achieve the desired target response. Many audiophiles assume that filter components such as capacitors, inductors and resistors “automatically” add distortion, but this is not true except for the most inexpensive parts. Extremely high quality parts will not add audible coloration or distortion, but the cost rises at an exponential rate when you’re buying great parts instead of inexpensive junk. For instance, we have seen 50-cent Chinese capacitors in speakers selling for several thousand dollars, while our new VR-35 Export Deluxe at $8,000 uses six pieces of $600 Teflon film-and-foil capacitors. We quickly discovered that the quality of our crossover design was paramount in the overall sound of the complete speaker system, so we spent more than 15 years refining the crossover circuits. This research included parts quality, circuit design, integration of the driver transient response, timbral response, and overall phase coherence. Although you can’t see the crossover inside a speaker cabinet, you should know that this “invisible” component will either “make or break” the sound of the speaker.
The last component of our design was to design a woofer system to extend the bass of our 6”-two way speaker down to 20Hz. At first, we used a 10” woofer in a sealed box, as we thought it would be simpler to design than a ported type of enclosure such as a bass reflex, transmission line, or passive radiator design. Although the sealed woofer cabinet sounded decent, it lacked the dynamics of other brands of speakers that used ports, so we decided not to “throw away” the internal pressure wave, but add it to the total bass response by experimenting with ports. Our first attempt at a bass reflex design was based on early Altec and JBL designs, which all sounded very boomy and had sluggish transient response. Clearly, this was a step backwards. After reading a British journal on the theory of Transmission Line speaker loading written in 1963, we realized that the resonant nature of a hollow resonator “bass reflex” cabinet was the cause of the boom and slow transient response. Further experiments proved that the transmission line’s greatest benefit was the reduction of the hollow cavity resonance (the cause of “boom”). Although it is somewhat true that very long labyrinth transmission lines enable the woofer to respond to deeper bass notes than a short line, the sheer size of a 12-foot labyrinth would be a “deal breaker” for most would-be owners. To reduce the physical size of the enclosure, we spent several years developing different types of chambers and designed special woofers to work within the parameters of our new design.
The original UK lines used long-hair wool as an internal damping material, but that attracted moths and was not as efficient as products developed a couple of decades later. We stuff our internal chambers with Dacron, which looks like cotton but does not settle at the bottom of the chambers and does not attract moths. The quantity and size of our internal chambers, along with the stuffing density of our transmission line designs is proprietary, but can be characterized by shorter internal labyrinths (made possible by developing specialized woofers for this type of application). Shorter lines enable even our smaller cabinets to have very deep, tight bass response. In addition, since our transmission line ports are large enough to fit your hand into, we have designed our lines to be adjustable for the quantity of bass power. This is mechanically done by the physical insertion of Dacron into the chambers to add additional resistance to the port output. If louder bass is desired, the user can reach into the port and remove Dacron, making the line less resistive and more efficient. We feel that being able to adjust the bass power of the speaker to the room is very important, since most rooms are not perfectly designed for a “perfect” speaker. A/B/X comparisons of our “quasi”-transmission line designs to other types of woofer systems have proven that our concept provides bass as deep and tight as any design on the market, regardless of cost, and the adjustability of our bass power is unique in the industry.
PD – What about current practice, do you continue to test with specific equipment or panels?
AVS – We listen to both digital and analog equipment. Since solid state amplifiers use no output transformers and can deliver constant voltage into a varying impedance load, we naturally use a wide variety of both analog and digital solid state amplifiers in our lab. This is because many of our speaker designs are rated at 4 to 6 ohms, and tube amps like 8-ohm speakers and above. However, since we believe that vacuum tube amplifiers have made a huge comeback and that many of our customers use tube amps, we have a variety of SET, push-pull pentode, and OTL tube amps in our lab as well. For listening to music for personal enjoyment, some of our various systems use VAC tube amps, KR “Kronzilla” amps, Audio Note SET’s, and several push-pull pentode amps made by VTL, Audio Research, and in the low cost range, Prima Luna and Audio Space amps. My personal favorite amplifier is the Kronzilla, of which I have owned several pairs since 1997 when I discovered this amp from our European Distributors. However, we find that many amplifier designs are “converging” and we have some solid state amps that sound like tubes and vice versa. In fact, when I was comparing our Pass Labs XA100.5 solid state amp (a single-ended transistor amp with zero feedback) to a VAC Phi Beta tube amp, which uses negative feedback and push-pull operation, one of my friends had confused the two amplifiers, thinking that the “tighter” VAC Phi Beta was the transistor amp.
PD – If you stand directly behind any VR series speaker, you will see the famous ambience tweeter. How do you create balanced full range sound behind the speakers when there is just that tweeter back there?
AVS – During the development of the original VR-4 speaker, we rotated the speaker on a lazy susan, taking frequency responses every 10 degrees, 360-degrees around the speaker. We discovered that bass and midrange frequencies will wrap around the enclosure, if the baffle is narrow. The exception is the front tweeter – due to the small diameter and the short wave lengths reproduced by the small diaphragm, the highs tend to beam over a 90-degree pattern and finally stop radiating spherical waves at 180-degrees. By using another tweeter at the rear, fed by a specialized ambience extraction circuit, the front and rear tweeters fill out the quasi-omnidirectional radiation pattern that is the inverse of the omnidirectional recording microphone.
I discovered through a series of psycho-acoustic experiments that listeners tended to prefer a speaker that sounded “open” compared to narrow, directional speaker designs. This finding was not all that surprising to me, since the goal of a loudspeaker is to attempt to mimic the recording, after all, and omnidirectional mics are used for the great majority of recordings. If omni mics we’re any good at picking up the full acoustic and spatial response, engineers wouldn’t be using them – they would use directional mics. However, since I have made recordings using both types, I can personally assure you that directional mics will not let you hear the “hall” sound and don’t sound close to the spatiality of a good omni mic.
Designing the quasi-omnidirectional radiation pattern was not an accident, it was done purposely to recreate what the mic heard at the original recording venue. After all, every engineering experiment has a target goal, based on either a theory or a wild guess, and by using a series of carefully designed experiments, the goal can be either proved or disproved.
PD – How much can you design the room out of the equation when crafting a new speaker? Given all the variables involved between rooms and equipment, how does a speaker designer start to make choices and what are the various weightings on key variables?
AVS – I am still shocked at how many “experts” state that since rooms are “bad” (i.e. reflective), a good speaker should be highly directional. What these fellows don’t understand is that a piano or other instrument isn’t directional and sounds fantastic in most any room. If you were to block off the sound of the piano, using baffles to direct the sound directly at both of your ears, you would hear a gross misrepresentation of a piano that has no basis in reality. In addition, concert hall design takes advantage of room reflections in order to enable the orchestra to sound full and rich, so your own listening room can be treated very slightly to reduce any deleterious effects that slight reverb might cause to the timbral response of the speaker system.
We use three different rooms with varying acoustics and dimensions, an “Average Small Room,” an “Average Medium Room,” and an “Average Large Room.” We can change the furnishings, rugs, wall treatment and so forth, in order to change the acoustics of the room. However, we start the design in the computer and design the various sub-assemblies by simulation (remember my aerospace background at Cal Tech). Then we test the concept in an anechoic chamber. Last, we do the final tweaking by ear in the three different rooms, making sure that the speaker will work well in any of the rooms. Even though we designed our UniField Model Three in the Average Small Room, all five of the reviewers, including Jonathan Valin of The Absolute Sound, wrote that the UniField Model Three sounded fantastic in any sized room. This is due to our “averaging” of the three different rooms’ acoustic response to the speaker’s radiation pattern.
PD – What are the long-term plans for VSA and what is your role in VSA going forward?
AVS -We will continue our quest to find the best possible methods to design the speaker with the highest value and sonic quotient on the market, especially in this down market. However, we are also working on a “Best Speaker In The World” type of thing that does not use conventional cone drivers. We own the exclusive license to the FPS transducer invented in Japan, which uses 200 motors (voice coils and neodymium magnets) to drive a 12” square flat plastic panel in “push-pull” operation. This driver has the fastest, most accurate impulse response ever recorded and has the potential bandwidth (with augmentation) to reach 10Hz to 100kHz. This is the first transducer I’ve ever heard that has zero coloration or distortion, it’s pretty much an example of Japanese perfection.
The transducers are hideously expensive, so expect a price tag of around $250,000/pr for a speaker the size of our VR-11SE. It should have a sensitivity of close to 100dB, impedance of 16 ohms for tubes and a special 8 ohm version for solid state. It WILL have the highest level of dynamic range and fidelity not heard in any other speaker design, period. We don’t intend on selling many of these flagships, but we want to prove that we’re still on top of our game, with many exciting developments just around the corner.