The Traveling Salesman

Ed Lim, properly known as Eduardo B.E. de Lima is a very unconventional thinker. He’s also the designer of his own line of pint-sized tube and hybrid amps and preamps plus matching speakers. They operate according to principles that he’s researched and discovered after more than five years of solitary experimentation, pursuing along the way wild and wondrous paths of self-perpetuating investigations quite outside the box. Jennifer Crock, an altogether far-out thinkeress in her own right and creator and purveyor of the Jena Labs cables and contributing writer to Positive Feedback, has called Eduardo one of the most brilliant audio designers around. How come you’ve never heard of him? Well, that’s precisely the point of this article. Until recently, I was oblivious to Eduardo myself, so don’t feel too alone in your ignorance.

In his native Brazil, Eduardo (shown in the picture left with Srajan Ebaen) has marketed and sold his products under the Audiopax brand. If this name suggests making peace with your audio system or feeling peaceful about it -- as in no-longer-compelled-to-look-elsewhere -- this is surely as intentional as are the ridiculously tiny transformers, the unbelievable sound and the extremely affordable pricing -- say $3500 USD for preamp, amp and speakers as currently sold in Brazil. But I’m getting ahead of myself.

I first met Eduardo in that arid zone of no-man’s land called City of Lost Wages, Las Vegas for short. It’s CES 2000. Assessing my dazed looks and sunken-in cheeks, the folks from Sound Application previously introduced in these pages as makers of the CF-X conditioner, determine that I’m overdue for a vital tune-up. They drag me with little resistance to authentic Chinese food somewhere way off The Strip. A very nice fellow is along for the ride. The healthy chow leads to intelligent conversation, giving support to the notion that you are what you eat, at least in your own mind. With my curiosity piqued by the unfolding stories, a stopover at Eduardo’s motel apartment afterwards becomes de rigeur. In these very humble and quintessentially unpretentious surroundings, the man has set up his system. Note the preceding words: set up. Looking at his jerry-rigged cabling -- simple home-brew tri-braids sans shielding but with some very funky RCAs, generic 12 AWG copper speaker cable most likely right from off the spool at Radio Shack -- I feel as though teleported into the impromptu lab of a traveling audio salesman. Boy was I about to be filched -- robbed of my preconceptions, that is.

Despite its unassuming cables and the dusty old Parasound CD changer up front, this system turned out to be the proverbial jaw-dropper. Within the first few notes, the fabulous immediacy and purity of single-ended triodes spread like a magic aroma through the room and proved the soothing antithesis to the solid-state glare and home-theater madness I had just suffered at the Hilton exhibits. However, I had also heard some mighty fine systems in the Alexis Park and St. Tropez high-end displays, and this bonsai-sized impostor was in no way embarrassed by those gold-studded aural memories. Something was going on here, and Eduardo’s unusual story begged to be told. American audiences hadn’t been exposed to this type of sound for so little. Neither had I for that matter. The pricing, in fact, seemed downright silly. In short, all the makings for another Y-Files column presented themselves very elegantly and quite on their own accord. With thanks to James and Connie Weil for the introduction, here we go.

The failed sales pitch

When you think single-ended, what comes to mind? Don’t be cute and quip nothing to make yourself sound like a Zen master. It’s, of course, true that a bit of single-ended-ness injected into our mental lives would drastically improve upon our clarity, focus and ability to manifest what we truly desire. But if it’s good sound you’re after, contemplating emptiness or your breath alone won’t do. Consider the following instead, and let’s concentrate on facts, not fiction.

• The basic SET circuit architecture is extremely simple and requires a minimum of parts.
• Most SET amps eschew global negative feedback.
• By design, SETs don’t require a phase splitter.
• SETs always operate in class A
• Unlike push-pull output transformers, an SET transformer usually has an air gap that creates a telltale kind of distortion spectrum.
• SETs produce only low-order distortion, predominantly second-order, with some third-order at lower levels and higher order ones only at the outer edge of their power band; but the distortion they do produce is significantly higher than that of other circuit topologies used in comparable applications.
• SETs have high output impedances, often up to 4 ohms.

Most designers even of solid-state amplifiers can graciously admit that the inherent simplicity, low parts count, lack of feedback and class-A operation of SETs go a long way towards supporting speculations why, as a whole, SET amps do tend to sound very good at least in the midrange. You see, even detractors can revel in the magnanimous but momentary gesture of praising the enemy. They know full well that at the very next opportunity, their practiced repartee will create serious injury. Why not look good while you’re preparing to smash into a bloody pulp he who with 7 to 10Wpc is clearly the least powerful of your opponents and sans feedback also the most defenseless?

Continue whenever you’re ready to be demolished. Simplicity can be achieved by other means and without having to contend with high distortion and an output impedance that tends to interact with a speaker’s impedance behavior and modulates its frequency response often to the tune of a few dBs.

SET aficionados now routinely get flustered. Without realizing their imminent defeat but sensitive to running out of ammunition, they counter with faux bravado that proper care in speaker matching -- i.e., selecting speakers whose impedance curve remains within a very narrow window -- overcomes the latter. But the unstoppable protagonists now unerringly go straight for the jugular, announce checkmate and point out that lovers of the SET sound are merely into the euphonic effects of octave-doubling -- second order distortion -- and thus prefer distortion, even distortion of the admittedly benign kind, to no distortion at all.

And this, my friends, is usually the end of that discussion. There simply exists no comeback, elegantly prepared or clumsily improvised, that overcomes this last and final assault. How to explain the very odd fact that many very sophisticated listeners prefer the single-ended, zero-feedback triode sound despite the concomitant high distortion spectrum? Can distortion, no matter what sort, truly be a good thing?

The comeback

This very question plagued Eduardo de Lima five years ago when he looked in puzzlement at a primitive SET kit he just completed. After years of building one-up amplifiers and speakers, he had finally ordered the requisite parts to build his first classic SET amplifier. The Brazilian and American hi-fi press of late had repeatedly covered such designs with much praise for their sonic performance. Working as a telecommunications engineer by day and thus thoroughly trained in scientific procedure, de Lima wanted an up-close-and-personal look at this antiquated topology. Surely the press had to be quite in error with their pronouncements. Nothing that a rainy afternoon with a soldering gun couldn’t quickly bury as reviewer’s nonsense, though. So Eduardo fully expected his SE encounter to be brief and satisfyingly disappointing. Like others before him, he found out that his expectations were to receive a massive high-voltage shock. This archaic circuit produced by far the best midrange he had ever encountered, easily besting his old Quad II on some homemade speakers that featured a KEF bass driver, a Peerless midrange and a JVC ribbon tweeter.

Things now could have ended there, with yet another hobbyist finding aural Nirvana, perhaps converting a few acquaintances and building a handful of amateur amps in the closet for the closest friends amongst them. For the sake of my story, that’s of course not what happened. Instead, de Lima looked at this unexpected result as a challenge. The glove had been thrown into his path because of his own curiosity. Being responsible for his actions, he promptly bent to pick the glove up and then set out trying to grasp why what shouldn’t work worked so damn well. He was also intent on divining whether the weaknesses in bass, treble and dynamics his kit amplifier displayed could be improved upon to match that glorious midrange. This is where things get unconventional and the real story begins.

By virtue of his scientific training, Eduardo is thoroughly grounded in reason. He’s not a mystic who accepts happenstance at face value. This forced him to stand firm by his conviction that an increase in plainly measured distortion is absolutely counter-intuitive to enhanced enjoyment of music. It had to be wrong; never mind that it sounded absolutely stunning. He was probably looking at distortion in a skewed way, not comprehending how it actually worked in the real world. There’s measured distortion as simplistically expressed by Total Harmonic Distortion (THD). Then there’s distortion as actually perceived by the ear. de Lima began to suspect that maybe one was quite different from the other.

It takes a particular kind of individual to be confronted by mystery and commit oneself to solving it. Most of us opt for exploiting those particular effects that are empirically known. Maybe we even take credit for the results, but in our hearts we know full well that we don’t have a clue how and why these things work as they do. I’m certain that my brief exposé can do but little justice to the amount of research and original thinking that had to occur before what you’re about to read -- nonchalantly and in a few brief minutes -- could be assessed, verified and put into practice by this soft-spoken Brazilian music lover and engineer. Consequently, his pride does not center so much on the actual Audiopax products per se. Rather, he perceives as his greater achievement the creation of the theories that become validated in the actual performance of the products they made possible. When famous designers like Neil Sinclair of Theta show undisguised surprise and admiration for the sound achieved with so little, Eduardo de Lima has no doubts that his theories are right on the money. This understanding, scientifically expressed in graphs, circuits and equations, is his real contribution to the art of SE design. Of course, many a SET amp has been built that sounds good -- he’s the first to admit that. But his amplifiers not merely sound excellent because of happenstance, adherence to old, well-published circuits or a fortuitous pairing with simpatico loudspeakers. They perform well because they were intentionally designed to do so.

This should stop you in your tracks for a moment.

It seems I’m blatantly stating the obvious while simultaneously insinuating that most engineers don’t design for good sound, doesn’t it? I admit I’m dramatizing a bit for effect, but I can assure you that SET designers on a whole do not pretend to fully understand why exactly SET amps do what they do in spite of all their apparent failings. Rather, they tend to follow empirical precedents and fine-tune proven circuitry with better parts. The more adventurous amongst them actually attempt to make SETs behave more and more like solid state. This has given us SETs with negligible to no output impedance, wide bandwidth, low distortion and, some claim, sonics that are quite a departure from classic triode sound. The very appearance of such designs might proclaim that what traditional SETs do is intrinsically wrong and must be corrected by cloning the rightness of the solid-state camp. Even the opposite has been attempted. Nelson Pass of Pass Labs designed the Aleph series of single-ended solid-state amps, now superseded by a new topology, to make solid state act SET-like and thereby exhibit some of their famous purity. Bob Carver’s Sunfire amp, still very much current, sports a current source output that raises the standard output impedance to approach that of a SET. All these are indications that while SET goodness is acknowledged, nobody really understands why it works when it does. However, there’s a reverse certainty about why things don’t work when they don’t.

As any scientist is want to, de Lima kept challenging his budding theories to ascertain where adjustments were required or whether assumptions had cropped up that would prove erroneous in a different context. If the typical zero-feedback triode sound was primarily due to distortion characteristics and their interface with real-world loudspeakers, a properly designed solid-state amplifier with the right distortion behavior should be able to at least come sonically close. de Lima’s hybrid amp with MOSFET output stage does indeed resemble the sound of his all-tube unit to an uncanny extent. I’ve heard both amps side by side. I’m comfortable saying that while not absolutely identical, the similarities were much closer than I ever thought possible -- another item that set off my internal alarm to note that something very unusual was going on in that motel room in Las Vegas. To make matters more compelling, de Lima has applied his understanding of distortion to investigate different tube types and claims so that, with proper implementation of circuitry and transformers, a 6L6 can be made to sound remarkably similar to a KT88 or a 300B. This is downright heretical stuff! It would be awfully hard to continue selling blue-blooded $1000-per-pair Western Electric 300Bs if this South American curandero could achieve comparable results with a commoner’s tube like a Ruby Tube 6L6. Fortunately for some, de Lima has relinquished trying to clone 300Bs and shifted his focus on the amplifier-speaker interface instead.

So what’s going on there that aroused his curiosity?

The close

It’s rather obvious but not readily acknowledged that the electro-mechanical moving-coil loudspeaker is the one component that exhibits by far the highest THD measurements of any devices in a CD-based system. Amplifiers and digital components generally don’t suffer from any low-order distortion whatsoever. Their deviations from perfection occur as minute but dissonant amounts of higher-order artifacts. Traditional moving-coil loudspeakers produce fair amounts of second-order distortion plus lesser levels of third harmonics at lower levels. But they do not display any higher-order artifacts unless pushed to the very edge of their power band.

If you remember what was said earlier about inherent SET distortion, you will have noticed that the total-harmonic-distortion product of an average moving-coil loudspeaker mimics that of a SET -- heavy on second, light on third and no appetite at all for higher-order distortion. What happens if two such devices, a SET and a loudspeaker, are connected? Will their individual distortions neatly sum or worse, multiply for even grosser deviations from harmonic purity?

For argument’s sake, let’s say that both devices produce 1% of second-order distortion. If the phase relationship between the fundamental and the harmonic is exactly identical in both devices, the distortion will squarely double to give 2% at the speaker’s output. One need not be a quantum physicist to strongly suspect that the chances for total harmonic distortion being identical relative to phase and over the entire frequency band, for both amplifier and speaker, are next to nil. In other words, it is extremely unlikely that the distortion of two such devices will cleanly double by addition.

I will spare you the mathematical-probability equations that Eduardo de Lima has produced at great lengths to investigate the complex interaction of subtractive and additive distortion behavior. Here’s the essence of his observations instead. At phase differentials between fundamental and harmonic from 0° to 120°, the total second harmonic distortion of both components, amplifier plus speaker, will continue to decrease, from simply doubling at perfect overlay [1+1=2] to partially subtract [1+1=1]. By the time the phase angle is at 150°, the total distortion will be half that of each component on its own, 1+1=0.5. At 180°, perfectly out of phase, the distortion will perfectly cancel to arrive at 1+1=0. Eduardo has further calculated that the chances for either an increase or decrease of distortion aren’t entirely arbitrary but governed by a two-thirds rule. If a speaker/SET amplifier combination gives more total distortion at the speaker’s output than either device produces on its own, reversing the polarity of the speaker/amplifier interface will most of the time not only reduce distortion but reduce it below the level the speaker generates by itself. It’s important to remember that this reversal of polarity must take place at the amplifier/speaker interface. Of course, doing so affects all frequencies and their harmonics, not merely those apt to be canceled.

These are simplified statements to be sure. To make them work requires intimate insight into the real complexities at work. For example, harmonic distortion varies both with frequency and amplitude. Negative feedback and multi-driver complexities in loudspeakers confound the issue even further. Nonetheless, it should be clear that the verisimilitude of the distortion pattern between SET amps and simple two-way dynamic moving-coil loudspeakers can be exploited. If understood properly, overall second-order distortion can be reduced where it matters -- not on the test bench but at the listener’s ear! As long as loudspeakers produce significant low-order harmonic distortion, low harmonic distortion sound in the room can only be achieved with complementary matching distortion in the amplifier. The magic word is matching.

Let’s put this in different, far more outrageous terms. This statement is very potent, so let it drop anchor properly: Amplifier distortion specs of vanishing low THD values are counter-productive unless such amplifiers are mated to (usually very expensive) speakers that possess ultra-low THD specs themselves. In other words, if you’re short on funds, you’ll get more natural sound by deliberately buying components with low-order distortion specs that cancel each other out when mated properly. You realize what this means for SET aficionados with simple two-way loudspeakers, don’t you? Second-order distortion rules!

Are you buying yet?

Hallelujah, I knew I wasn’t deaf when my ears cottoned onto the single-ended sound in a big way. What a relief. Of course, distortion is ultimately good only if consciously accounted for and reverse engineered back from the loudspeaker where it’s an intrinsic quality due to basic mechanical limitations. It must be deliberately designed into the amplifier in such a way as to reduce the loudspeaker’s inherent flaws by subtraction. Otherwise the entire enterprise remains an arbitrary, hit-n-miss affair effective only over isolated frequency bands. This is something most SET owners looking for just the right speaker can readily attest to. In his amplifiers, de Lima claims to have done exactly that -- match their distortion characteristics to those of the speaker. This, of course, required not just any but a particular speaker for maximum effect and so prompted him to begin design on his own. He further claims that by implementing the proper techniques, the distortion-reducing effects between SET amps and simple two-way loudspeakers, occurring naturally only in the midrange, can be extended into the bass and treble regions as well, areas where SETs traditionally suffer from poor performance.

Remember how one of the universally acknowledged evil aspects of SET goodness is their high output impedance? Eduardo knew from personal investigation that high source impedance tends to negatively alter the bass alignment of vented loudspeakers by introducing an impedance hump. Since he couldn’t eliminate this typical SET behavior without interfering with those other qualities he had already learned to view as assets rather than liabilities, he decided that his own loudspeaker would not only account for but actually exploit this high output impedance for improved performance. In other words, de Lima wanted to design a system where each component balances the distortion components of the others to achieve maximum subtraction at the listener’s ear. This approach is intrinsically holistic, like Oriental medicine. It looks at the whole picture rather than just the parts, as does Western medication, which is often intrusive and punishes with long-term side effects while temporarily impressing with potent short-term results.

Besides just generating the proper distortion, de Lima’s amp would have to meet his speaker yet a bit further still. In the process, it would undergo a massive size and weight loss. He assessed that the primary inductance of the output transformer -- a usually copious amount of coiled wire around a massive iron core, otherwise known as an inductor -- should be significantly reduced to help ameliorate some of the effects of the high output impedance. This would enable the use of much smaller transformers with better low-end power response, superior high-frequency extension and reduced cost. Now we’re getting into some very technical parameters, many of which are non-disclosure material. de Lima’s tiny custom transformers won’t see patents pending anytime soon because this would entail exposure of construction details. Remaining mute is much better and cheaper protection. Also, their manufacturer is a close personal friend of advanced age who won’t give away the secrets of winding geometries and core treatments as a matter of personal integrity. Suffice to say, then, that those curious enough can, at their leisure, peruse two lengthy articles by de Lima, published in Glass Audio issue 3/97 and 6/97 that deal with the accompanying equations and calculations. Frankly, they find me lacking in both ability and inclination to truly fathom.

Besides the use of unusually small transformers, de Lima’s tubes aren’t connected in the traditional triode or ultralinear mode either. He was quick to draw me his LM3 circuit diagram right onto a green-tea-stained paper napkin. However, I might as well have stared at ant shit, rolled up my pupils and predicted the future. I honestly couldn’t fully grasp what he was trying to explain about his "low mu triode with higher raw efficiency emulation" circuit. The best way to describe it in words seems to be "ultralinear mode with partial cathode coupling and a certain amount of cathode degeneration." It’s said to allow unusual control over the levels and spectrum of distortion and also benignly affect the output impedance. Of course, de Lima's bass driver turned out to be anything but normal either. He heavily modifies it himself because a woofer with his special requirements isn’t available off the shelf. I didn’t dare ask details. It’s obvious that this too falls under the proprietary category of don’t ask, don’t tell.

My intent here clearly is to introduce you to a serious thinker who has a very significant contribution to make that adds to what’s already known or suspected about the single-ended-triode phenomenon. But I’m not opening hunting season onto his hard-earned and novel solutions so those unwilling to do their own homework can practice convenient misappropriation. This leaves us then with the actual products.

The currently Brazil-based Audiopax brand produces an entire system that South American music lovers can enjoy right now for the equivalent of $3500 USD -- the APX01 SE preamp (right), APX02SE amplifier and CX303 loudspeaker. There are more expensive models in the Audiopax range to be sure, but de Lima bluntly states that the bang for the buck stops with this system. Incremental improvements are possible but cost disproportionately more, so this particular rig is boss. Whether it will eventually be marketed in the US and abroad remains to be seen. From where I stand, the designer’s very desire to keep his products as affordable as possible -- something which is openly reflected in their tidy yet very kit-style appearance --could prove to be a stumbling block with the US dealers. They might prefer a more polished look and insist that they’d sell more if it cost more. This is part of the audiophile perversity that’s become commonplace in not just our industry. A manufacturer’s real customer is no longer the end-user but the dealer who becomes the actual gatekeeper for his products. Gratify the dealer with profits, advertising support and an easy sell and you have a chance to get through to the end-user. Concentrate instead on pleasing the "real" customer -- you -- with low retails, great performance and functional instead of designer cosmetics and you’re most likely deluded to expect success any time soon.

But good performance, or better yet, jaw-dropping performance, is -- or should be -- part of an easy sell. On that front and as we shall see shortly, de Lima's theoretical groundwork seems to have paid huge dividends. But before you credit me with introducing Audiopax to the US music lovers, let me state that Eduardo de Lima attended the admittedly esoteric and hence relatively invisible Vacuum State of the Art Conference and Show in Silverdale near Seattle as early as 1997 and 1998. After his first showing, visiting V&T News editor Lynn Olson had this to say: "…one example of thoughtful amp-speaker matching is the Audiopax system. It features single-ended triode amps with the output transformer inductance tailored to the speaker’s roll-off which forms a third-order electro-acoustic filter. The speaker and triode amp were designed as a system and voiced together. Good idea!…"

After the second VSAC show, the next-generation Audiopax system with now phase-matched distortion characteristics in addition to the "inductance filter" attracted many comments of the best-sound-of-show kind. The system I was exposed to in Vegas and subsequently had in my house represents de Lima’s latest thinking and incorporates all he currently knows about consciously induced partial distortion cancellation and optimization of the speaker/amp interface, implemented to sell at a very reasonable cost. Rather than being a close-ended system that absolutely has to be used in just this context, the amplifier and speaker on their own are said to be as good or better than competing products. Obviously, the built-in matching only occurs to maximum effect when amplifier and speakers are used as intended -- together. This is how I heard and used them, except for a brief sneak with different speakers to satisfy my curiosity. The amp’s output tubes are 6L6s with 12AX7 pre-drivers. The preamp uses one 6SN7 per channel and the speakers sport an inverted Focal tweeter and a heavily modified, uncommonly stiff 7" bass driver with a tri-pleated surround. It's not unlike those found in the excellent French Triangle speakers and many professional models.

A compelling aural argument

Other contenders at the $2000 price point for preamp/amp or integrated SET solutions include offerings by Antique Sound Lab, Audion, Cary, Cary’s Audio Electronics Division, Unison Research, Welborne Labs, Wright and the X/Y combo by FI. I’m not aware of a single system, though, that includes purpose-designed speakers from the same maker.

Compared to the $3300 Audiomat Prélude I had on-hand plus my significantly dearer Art Audio Jota, the unassuming Audiopax, when used as a system, cruises right past the French push/pull integrated in terms of speed, dynamics and openness, coming closer to the Jota than one should rightfully expect. It’s very robust and meaty, throws a gigantic soundstage entirely out of proportion with its diminutive dimensions and positively trounces the Prélude in spatial detail retrieval and ambient recovery. These two latter qualities are in fact some of the very best I’ve every heard. The slight thickness and concomitant appearance of "drag" I hear in the push/pull design are entirely absent in this single-ended system. Using my resident Soliloquy 6.3s just for a reality check, the little pre/power combo exhibited a tendency for a certain whitish thinness, suggesting a less-than-ideal match. This offset the transparency/presence balance heavily in favor of transparency and lent credence to the designer’s insistence on a complete-system approach. This balance was immediately and brilliantly restored when the much cheaper CX303s were substituted. The Audiopax system does not emotively project as the Jota does. Emotive projection is my way of describing a system’s ability to have the music reach out to and for you through the air -- think of the personality of a master orator that tangibly projects intact across a large audience. The Audiopax presentation is different. It had me go to the music. To remain with the example, I understood the orator just as well, but this time I felt myself connecting with him rather than having him connect with me. Neither approach is right or wrong. They’re merely different and quite fascinating to explore.

The Audiopax sound is very refined and luxurious, ultra smooth and creamy. Yet it does retain the musical excitement that many SETs seem to relinquish in their quest for ultimate transparency and relaxation when the music just sits there but doesn’t get up and go. The Audiopax system does go, but without quite the Jota’s benign incisiveness in transient blister and microdynamic jump factor. It’s much closer than the heavier-set Prélude, though, which doesn’t really surprise me; push/pulls tend to be a bit fatter. The Audiopax is the Jota’s more gentle-spoken brother who slightly softens over some of the edges the senior lays bare in screaming Cuban trumpets, incendiary flamenco guitar strings or explosive rim shots. The system can be pushed past its limits in the bass at highly elevated levels, and both the Prélude and Jota have more testicular fortitude and continue well past where the Audiopax system stops. While intelligibility and nimbleness of bass are first-rate with the Audiopax system, the bigger players handily overtake it in the extension and heft department and, of course, are much less limited in the choices of speakers one can drive. As good as the Prélude is overall (its financially downscaled twin, the Arpège, received our coveted Reviewers' Choice recommendation and I do find the Prélude rather more satisfying than the very well-regarded Yakov Aronov Integrated), the Audiopax system more completely fulfills my personal list of priorities. In comparison to the Prélude and on my personal ballot, I’d have to declare it the clear victor. There is still appreciable reason why the Jota is more expensive, but the law of diminishing returns does rear its head. If your listening tastes don’t demand massive bass slam and the last word in attack, impact and scale, but favor refinement, serious spaciousness, purity of timbre, high resolution wedded to emotionalism, stunning soundstaging and zero-edge smoothness, the Audiopax system leaves nothing to be desired. It’s silly good.

Can you afford not to at least pay attention?

Anyone who currently listens to triode amplifiers ought to read up on Eduardo’s research papers -- look them up at http:/w/ww.audiopax.com or http://usuarios.uninet.com.br/~edelima. While digesting his findings, you should remember two things. Eduardo de Lima has been a member of the Audio Engineering Society since 1981. He presented his findings at the third regional AES Brazilian Convention in Sao Paulo during July 1999, where he publicly defended the continuing relevance of SET design in the face of transistor and non-linear amplifier topologies -- no slack challenge. Before all of this gets too technical for your taste, know also that Eduardo, during his MSEE program at Syracuse University in 1986/'87, worked as a research assistant in a project at the Belfer Audio Laboratories and Archive, which is part of SU. These archives contain a huge collection of old cylinders and records from the end of the last century through the '40s. The project at the time involved to recover as much of the sound of this collection as possible and included DSP processing on VAX computers as well as laser cylinder reproducers to avoid playing the cylinders with bamboo needles. This got de Lima involved in a lot of re-recording and helped balance his scientific bent with, quite literally, caring for music.

Reading his papers will convey a better understanding of some of the technical aspects that automatically take place each time you power up your SET system to enjoy the music. If nothing else, you’ll come away appreciating the simplicity of flicking a switch over feeling compelled to spend years crunching numbers, developing research models and building prototypes to validate ideas and growing theories.

I queried de Lima on whether DSP could achieve the distortion reduction occurring quite naturally in SET systems and which he exploits consciously and to a further extent in his own designs. The algorithms could certainly be written by anyone in possession of the necessary data on harmonic-distortion behavior. He warns, though, that the required computational brainpower to process such data in real time is still quite a few years out. Companies like Mark Schifter’s Perpetual Technologies are currently using DSP to combat distortion. This involves distortion correction in the time, frequency and phase domains. That in itself is a major feat and promises drastic and affordable improvements to the archaic and antediluvian mechanical device called a conventional loudspeaker -- more on that in a future installment. However, it does not address harmonic-distortion reduction. A brief e-mail exchange with Mark Schifter enlightened me to the fact that it took countless years of writing code before his first corrective product could see the light of day. Calculating corrective coefficients based on the impulse-response data for a particular speaker model still takes more than one hour of mainframe processing using their proprietary software. If you calculate 3600 second/one hour of processing into a 300mHz Pentium II, the awesome complexities of DSP-based distortion correction become apparent.

Simple does it

Harmonic distortion reduction at the listener’s ear remains a new frontier to be conquered that isn’t even on most audiophile maps yet. Eduardo de Lima asserts that a properly designed or assembled SET system can diminish second-order harmonic distortion quite effectively and at comparatively little cost. Many listeners have intuitively come to the same conclusion while struggling with the apparent dichotomy of how more can equal less -- how the famed SET euphonic second-order departure from reality can sound so bloody realistic. Now you know -- it’s a subtractive phenomenon. de Lima’s Audiopax system is further proof that certain automatic responses of SET systems can be deliberately optimized to improve the results such that to achieve them with other types of "more advanced" amplifiers would be infinitely more costly. Now contrast these statements to state-of-the-art digital signal processing, ultra-low-distortion amplifiers and superbly engineered flagship speakers. You arrive at one of those paradoxical slaps in the face. You could react annoyed, offended or simply consider it a wake-up call. I don’t know about you, but I’m saving myself a cold shower to head right over onto my listening chair. It faces my Art Audio single-ended Jota and some simple two-ways. Granted, I am a self-righteous fellow. That’s the German in me. I’ve worked hard on it all my life. But you get the point. Sometimes simple does it, even if it smacks of heresy. And a bit of heresy adds spice. Now where’s that smoky Brazilian samba album? I’m feeling pesky today. I’ve finally understood to appreciate the wisdom of my single-ended folly. I’m sane after all. Watch me as I light up another ultra-fine Czech triode and dissolve contentedly into the emotional splendor we call music.

And now for the really good news

While I was working on this article, things were apparently brewing at Audiopax. Shortly before I delivered my final copy to our editorial, Eduardo e-mailed me with excellent news for the US and Canadian music lovers. He is relocating to the San Francisco area to set up a new manufacturing facility and begin a marketing campaign for the Audiopax brand in America. It is now absolutely certain that US and Canadian music lovers so inclined soon can either call me a liar or a reasonably astute listener. You will be able to taste and perchance purchase for yourself Audiopax Eternam for a dull nickel or whatever is the opposite of a pretty penny. Eduardo has also signed a contract with Volkmar Drübbisch of pARTicular, introduced in these pages last month as looking for additional venues of artistic expression outside of uptown audio racks. Volkmar will provide his industrial design expertise to fashion a new contemporary and integrated look for the complete Audiopax system as described and photographed in this article. If this turn of event appears conspicuously coincidental and suggests that this columnist has been meddling in audio politics, I, like any other glib politician, shall only confess that whatever introductions may have been rendered were granted purely in the interest of the greater common good.

If the Audiopax system as pictured here should turn out to be what’s initially marketed in the US, I can see it happen through direct sales purely as a function of its present kit-style appearance. Don’t let that stop you. If superior sonics and musical communion are your top concerns, I really can’t think of a better system for the money. All compatibility concerns have been addressed to ensure that you’re hearing all that your components are capable of and exactly how the designer intended them to perform. Also, 6L6s, 12AX7s and 6SN7s are some of the most common of all tubes, in abundant supply, reliable and very inexpensive to replace. Should you wish to embark on a bit of tube rolling and NOS experimentation, you’re looking at a system total of six tubes of which there are plenty of variants. Forget biasing -- this is plug’n’play computer-age stuff disguised beyond recognition by its heavily retro thermionic and hobbyist appearance.

...Srajan Ebaen
srajan@soundstage.com

• The Y-Files Current Issue