I've positively reviewed tube amps that sounded wonderful to me, then performed poorly on JA's test bench. It was like having my pants pulled down in front of a large crowd of people. But Lamm Industries' impressive track record of measurements as published in Stereophile gives me confidence that this time my pants will stay up—even if low-powered, single-ended-triode (SET) amplifiers such as the ML3 Signature usually fare even worse than their push-pull relatives.

"Are you a technical person?"
That's what Vladimir Lamm, a Russian ÇmigrÇ and former rocket scientist, asked me as I prepared for this review. I told him that, yes, I am technically a person, but when it comes to amplifier circuitry, I am more conceptual than technical, and that I think that also describes most Stereophile readers.

So, conceptually, here's what you need to know about the ML3 Signature. It is a zero-global-feedback, single-ended-triode, class-A (meaning that a constant voltage, or bias, is applied to the tube's grid) monoblock amplifier. Although Lamm feels that using feedback is merely a Band-Aid, the ML3 does have two switches, offering 1.2dB and 2.4dB of negative feedback. These affect only the output stage, and serve to lower the output impedance. They're there, Lamm admits, just so you can hear for yourself how using feedback degrades the sound.

Four on the Floor
Vladimir Lamm's ML3 Signature monoblock amplifier with ML3-PS power supply is unlikely to win any awards for its industrial design—or lack thereof. If you like 1930s-style "lab-grade" basic black, you'll love their looks. I do. They look like something from Victor Frankenstein's laboratory.

Each chassis weighs about 80 lbs and measures 15.8" wide by 8.2" high by 20.2" deep—the four of them occupy a lot of floor space. Chances are, if you can spend $139,490—yes, that's the price—for a pair of 32W SET amps, you've got the space. As with any product with exposed vacuum tubes, use caution if you have kids, animals, or anything hanging above or on adjacent shelves.

The design of the ML3 centers around the care and feeding of its single, direct-heated (like a light bulb), GM-70 triode transmitter tube claimed to continuously output an unusually healthy (for a SET) 32W, up to a maximum of 37W. The ML3's other tubes are a single 12AX7/ECC83 and four dual-triode 6N30P-DR, each matched to its socket. At the chassis rear, behind the tubes, sit two formidable transformer cases.

Among other premium components, the ML3's internal parts include military-grade Dale metal-film resistors; Caddock power-film resistors; Cornell-Dubilier and United Chemi-Con electrolytic capacitors; and Electrocube, Elcon, and Roederstein film caps. Connectors are gold-plated Neutriks and Fischers. Double sets of biwirable speaker terminals for each of the three transformer output taps (4, 8, and 16 ohms) line up along the amplifier chassis rear. Lamm claims the ML3 delivers its full 32W output from all of these taps into matching loads. The input is single-ended RCA with a "utility" XLR connection included for convenience.

The ML3 is fed by an enormous outboard power supply, the ML3-PS. Lined up across the front top of the ML3-PS are six 12AX3 rectifier tubes; behind them are four big transformer cases, though rapping each of these with a knuckle indicated that only one is completely occupied. The ML3-PS has separate plate and filament transformers, the latter located in the ML3 chassis, six filter chokes, and, in the high-voltage power supply feeding the output stage, the "highest quality" film capacitors. The ML3-PS is connected to the ML3 via a multi-pin umbilical.

Considering the price, all of that is the least you should expect. But more important than any of it are the custom-wound output transformers that are key to any tube amp's ultimate sonic and measured performances, and especially for a SET. Those transformers are any tube amp's "special sauce," and among its most costly elements. The ML3-PS's plate, filament, and output transformers, all also custom wound, are mechanically isolated from both the transformer cover and the chassis, and suspended within a special vibration-absorbing material.

Associated Gear Critical
Even with its relatively healthy (for a SET) output, the ML3 must be paired with a sensitive speaker with a relatively benign impedance curve. I heard two ML3s driving Wilson Audio Specialties MAXX 3 speakers in a large hotel room at a Consumer Electronics Show, and came away less than impressed. The sound was anemic and lacked dynamic drive and slam, though tonally it was mesmerizingly lush and smooth.

Now that I've got a pair of Wilson's 92.6dB-sensitive Alexandria XLFs (I reviewed them in the January 2013 issue), I felt it was appropriate to review the ML3 with them, and even though I'm not a technical person, Lamm agreed. As JA pointed out in his measurements of the XLF, "Despite the Alexandria's imposing bulk, it will play at high levels with only a few watts. (During the in-room measurements, performed at a reasonably loud level, the darTZeel [NHB-458] amplifiers' meters never indicated more than 5W peak.). . . . [T]he speaker will not be a difficult load for the partnering amplifier to drive."

Setting up the amps was easy, and Lamm makes it convenient to measure the output-tube bias, which is specified at 1.1V DC—but for $140k, he probably should supply a good voltmeter instead of an electronic parts catalog showing a good one to buy! I've got a good one, and out of the box, the bias of both amps was correctly set at 1.1V. Over the course of the review it eventually dipped to 1V for both amps and had to be rebiased, but otherwise, the ML3s presented no problems and needed no maintenance.

Lamm suggests replacing the 12 half-wave rectifier tubes in the ML3-PS every 18 to 24 months (about $120 total) and the ML3s' GM-70s every 18 months. He doesn't specify the cost of a matched set from Lamm Industries, but again, if you're spending this much for a pair of amps, you're probably not all that concerned.

The Stage Is Set for Sound!
But first, some perspective. I sat a few months ago with my wife, Art Dudley and his wife and daughter, and some others in the rooftop ballroom of Manhattan's St. Regis Hotel, where we were treated to the Emerson String Quartet and an additional cellist performing Schubert's Quintet in C Major, D.956. The acoustics were excellent, and our table was close to the stage. As I listened, I tried to analyze the differences between what I was hearing and the sound of the best string-quartet recording I've got.

First, anyone who tells you that the sound of any audio system even approaches what you hear when you're in a room in which musicians are playing is fooling themselves and you. (Of course, we like to be fooled.) The dimensional and spatial cues on recordings are wrong. Attacks in even the best recordings are ham-fisted and lack textural grace and complexity. Dynamics, particularly microdynamics, are stifled, and harmonics are either shortchanged or overemphasized. So no matter what, you're not going to get "live" sound at home.

The best solid-state amps always measure better than the best tube amps, produce greater transient speed and clarity, and generally produce far greater dynamic contrasts, and superior bass extension and control. Generally, solid-state amps deliver levels of transparency that few, if any, tube amps can manage.

However, compared to the best tube designs, even the best solid-state amps produce an overhyped transient attack not heard live and can sound harmonically drab. And often, solid-state's superior transparency is accompanied by an analytical and "crunchy" sonic aftertaste, with the leading edges dominating, and the harmonic envelope licked and closed.

Tube amps, on the other hand, are generally harmonically overripe, sound richer than life, slower and softer than life on transients, and lack solid control on the bottom. They tend to be noisier, bloomier, and "soggier" than life.

So where did the Lamm ML3 fall? In some ways, predictably within the aforementioned tube parameters, and in some ways well outside them, in a realm of its own. Thus, when it comes to gushing about the ML3's sound, count me in-within limits.

Yes, the ML3 sounded lusher and bloomier than life, with a midrange that almost landed in my lap, but it never failed to create a warm sensation in my chest that flowed up to the brain and lingered there for as long as the tubes glowed. And while the ML3 erred slightly on the side of bloom, it did so while also presenting the most transparent and open midrange I've ever heard from a tube amp—midband richness and transparency, produced in large part because this amp, despite its midband richness, was notably quiet.

Most startling, though, was the ML3s' effortlessly clean and airy high-frequency extension. The Lamms were ruthlessly revealing of differences among speaker cables—even The Pathetic Randi would hear them. Choose the wrong cable and the amp could sound hard and bright—or slightly clogged in the upper midrange and meager and overdamped on bottom. Choose the right cable, and it was smooth sailing from top to bottom.

Speaking of bottom, the ML3 may not have had solid-state's fast, lean low-end attack, which is needed for the correct reproduction of amplified bass, but the extension was there and attacks were fast, clean, and nimble enough to produce a more-than-adequate foundation for even hard rock. Not the best there is, but better than expected.

While the ML3 produced the rich harmonic palette expected from tubes, the naturalness of its portrayal of instrumental attacks put it as close to what I heard in the St. Regis ballroom as I've heard from any amplifier.

The familiar vinyl editions of Bill Evans's Waltz for Debby (45rpm LPs, Riverside/Analogue Productions 9) and Live at the Village Vanguard (45rpm LPs, Analogue Productions) produced floor-to-ceiling, wall-to-wall spaciousness, and in that space appeared convincingly three-dimensional images of instruments that weren't tubey or diffuse. With the right speaker cables—in this case, David Salz's latest, the Wireworld Platinum Eclipse 7—the late Paul Motian's shimmering, ringing cymbals sounded neither tube soft nor solid-state hard. Scott LaFaro's double bass sounded as nimble and natural as I've heard from these records. The attack, sustain, and decay of all three instruments was as graceful and lifelike as I've ever heard them, particularly the generosity of the sustain, which created (just so you don't think I'm getting carried away) an admittedly somewhat impoverished version of the floating, lifting sensation you hear live.

Not to be cynical, but I've heard, over the years, countless variations on that sentiment. Not to be naïve, but it rang with somewhat-greater-than-usual sincerity when given voice by 45-year-old Richard Wugang—founder, with his late father, of Virginia-based Sophia Electric, Inc.

Wugang speaks with similar import of his audio upbringing: of the days he spent helping his engineer father disassemble and study some of the classic audio transformers that came their way, the older man noting in a green ledger the results of their experiments. Years later, after Richard had achieved success as a banker, there came a day when he looked up from his desk to see his father in the doorway, the old, green ledger in hand, asking to join him in starting an audio company. Neither looked back.

In 2001—a year not known for coaxing tears of wistful nostalgia from the eyes of audio manufacturers—Sophia Electric began selling new mesh-plate 300B triode tubes. One year after that, Sophia introduced its first commercial-production amplifier: a US-made, single-ended, 300B design heavily influenced by the classic Western Electric No.91-A amplifier, multiple samples of which have been owned by both father and son. Today, this 300B amp endures in the Sophia line, in three different guises: the 91-01 monoblock ($4500/pair); the 91-03 stereo amplifier ($6500), which features higher-quality parts; and the 91-05 stereo amp ($10,000), which offers Sophia Electric's finest-quality parts, chassis, and transformers.

Sophia Electric's business model, in which audio amplifiers and component parts are sold via the Internet (home trials are available, the only requirement being for the customer to state that intent at the outset), doesn't satisfy the "five-dealer" rule regarding eligibility for a Stereophile review (footnote 1). Yet because that policy is motivated by the desire to focus our efforts on products from real companies—thus sparing readers the hassle of buying, on our recommendation, products that mightn't be supported a few years hence—I was swayed by Sophia Electric's longevity to take a look at their level one Heritage Series 300B amplifier.

Description
The Sophia Electric 91-01 300B is a single-ended power amplifier designed around the 300B directly heated triode tube, operated without negative feedback. Its input/driver tube is a Chinese-made 6SN7 dual-triode, implemented not in the rather common series-regulated, push-pull (SRPP) mode, but rather, according to Richard Wugang, in a manner that "allows enough voltage swing for the 300B to deliver 8W." According to Wugang, an SRPP arrangement "has some advantages, but it sounds too aggressive." The precise details of the 91-01's input circuit were kept from me by a smallish input-circuit board, containing four capacitors and seven resistors, that obscured from view the socket for the 6SN7 and much of the pertinent wiring.

The input stage is capacitor-coupled to the signal grid of one of Sophia's own Princess Mesh Plate 300B tube ($475/pair when sold separately). Rather than applying a fixed bias voltage to the grid, Sophia runs the 300B in autobias mode, the cathode held approximately 70V above ground. Thus, despite a highish 425V rail, the output tube sees considerably less than that between cathode and plate. Separate hum-reducing potentiometers, accessible from above the 91-01's top plate, are included in the heater circuits for the input and output tubes. (Sophia guarantees their output tubes for 1 year; Richard Wugang says that one can typically expect an output tube in one of his 300B amplifiers to last from five to seven years.)

The 91-01's power supply is built around a Chinese-made 5Z3PA full-wave rectifier tube, and the rail is smoothed by a pi filter of the usual sort, involving an unusually hefty choke, apparently made in-house. Filament voltages are all AC, supplied by dedicated secondary windings on the mains transformer, itself also a Sophia original.

The output transformer, too, is designed and made in-house—a point of some pride, especially given the founders' years of research. Richard Wugang says it's wound on a standard IE core ("so we do not lose midrange magic"), using a mixture of vintage and "high-tech" core materials. According to Wugang, he and his father determined that, contrary to the prevailing attitude among designers and hobbyists, a smaller-than-average core produces a better-sounding transformer—and he added that Sophia has "found a unique way to get wider bandwidth."

All of this is built into a steel chassis, painted black inside and out, with a separate interior subplate for the tube sockets and separate steel covers for the mains and output transformers. Apart from the input-circuit board, described above, all wiring is point-to-point and routed somewhat haphazardly. Indeed, while I greatly admired the ruggedness of the little chassis, and the transformer quality is presumed, through listening, to be superb (see below), I was a bit disappointed by the 91-01's build quality: A few solder joins between component leads were sloppy, the large choke was made to fit the chassis by bending its frame, and, most concerning of all, the leads of some high-voltage capacitors were overlong and underinsulated.

Setup and installation
Installation of the Sophia Electric 91-01 300Bs was straightforward: Separate loudspeaker terminals are provided for 4- and 8-ohm loads, and I used the latter with both the DeVore Fidelity Orangutan O/96 and my Altec Valencia speakers. I did not, during normal use with their standard output tubes, need to adjust the Sophias' hum pots. I did, however, have to make slight adjustments when I swapped in my own pair of Western Electric 300Bs, owing, no doubt, to the latters' very different current-draw characteristics. No big deal.

As with all such products, the 91-01 owner is advised to use these amps on a solid shelf—never a carpeted surface—in order to promote cooling. I complied, with no additional attention paid to mechanical isolation. The Sophias ran warm but not unduly so, although the smell of fresh paint was a bit of a nuisance after the amps had reached their operating temperature.

The sketchy owner's manual neglects to mention anything about the function of one control: a top-mounted toggle switch that turned out to be an AC ground float. (Its use proved unnecessary in my system.) The use of the hum pots was also, I thought, given short shrift—although I was heartened that the manual warns the owner against doing such "silly things" as taking the 91-01 into a car wash or cleaning its tubes in the dishwasher.

Listening
Notwithstanding the manner in which the review pair matured and "opened up" over several weeks of use, the Sophia 91-01 300B impressed me from the start with its smooth, rich, warm, very well-balanced, altogether musical sound. It had decent color, very good (if sometimes a bit exaggerated) texture, and an acceptable sense of touch and force. From the moment I installed the Sophias in my system, I enjoyed them thoroughly, and though they didn't sound nearly as refined or as eerily human as my far more expensive Shindo Corton-Charlemagne monoblocks, I was never, during the review period, as anxious to return to my references as I usually am while reviewing strange gear.

With the musically brilliant and very well-recorded 14 Bluegrass Instrumentals, by Country Cooking (LP, Rounder 006), the Sophias' top-to-bottom balance was just about perfect, and their slightly warmish character served well an album's worth of fare that can get a bit glary with some gear. That said, the sounds of the twin (!) banjos of Pete Wernick and Tony Trischka did at times betray just a touch of clatter, and Kenny Kosek's fiddle had a little too much texture: qualities that might be associated with higher-than-average levels of harmonic products.

Audio engineers have been doing those things for ages. Have they improved their craft to the same extent as the people who, over the same period of time, earned their livings making, say, automobiles and pharmaceuticals? I don't know. But if it were possible to spend an entire day driving a new car from 50 years ago, treating diabetes and erectile dysfunction with the treatments that were available 50 years ago, and listening to 50-year-old records on 50-year-old playback gear, the answer might seem more clear.

Actually, I was just kidding about the answer not being clear in the first place: Compared to the advancements achieved by their colleagues in other fields, audio engineers might as well have spent the past 50 years stripping the leaves off branches and dipping them into termite hills.

Certain new developments have been worthwhile. Hats off to everyone who makes low-friction tonearm bearings, transformers that resist saturation, and other marvels that eluded our elders. Modern capacitors work wonders in some applications, as do modern resistors. Silicon diodes, rectifiers, and regulators are useful. The KT120 is a nice tube. Modern plastics and adhesives have made possible some excellent panel-type loudspeakers. Through the miracle of science, we can now safely and effectively wash 80 years' worth of records.

But it seems the majority of engineers in today's audio industry put their greatest efforts behind the least worthy ideas. Complex cables and their integral "correction" systems. Systems that propagate soundwaves from behind and around the listener. Wireless this, remote that, gold-plated this, carbon-fiber that. Products in all categories that can scarcely be moved, let alone lifted, let alone afforded, simply because sheer bulk is the only way their inventors could imagine to make the things better. And, of course, ever-more-powerful amplifiers, as would be required to drive the industry's ever-less-drivable loudspeakers. God help us.

Remarkably, there remains, in the mainstream of perfectionist audio, a sticky film of reverence for Quad ESL speakers and Garrard 301 turntables that could nauseate at 20 paces—like hearing the members of Styx or Queensryche declare their love of Johnny Burnette and Gene Vincent. This seems especially true of the amplifier makers: "Marantz 8B? Greatest amp ever, and a profound influence on our work." Sure. That explains the Mercedes S-class prices, the ridiculously thick laser-cut faceplates, the complex, heavily regulated circuitry, and the output-power ratings that reach into the hundreds of watts and beyond.

That last one is especially hard for me to swallow.

The sounding board
For 18 years I've reported on that strange corner of the world where people insist on playing records through low-power amplifiers and high-efficiency loudspeakers—which, of course, is how the thing was done at the dawn of domestic audio. Ever the aspiring John Reed, I became a convert to the cause I covered: Thus I've not only spent a cat's age writing about scores of flea-watt amps and sensitive speakers, I've bought—and occasionally built—a goodly number of the things for myself.

Over time, I've become more unshakably convinced that this is the best approach for a record lover such as I, who values tone, touch, and musical flow over all else. (There are a lot of other all elses, from which you and every other listener are free to choose.) That conviction led to my purchase, last year, of a crazy, hulking pair of Altec 846A Valencia loudspeakers. Before their arrival, I had never enjoyed such a high and wild level of system responsiveness in my home.

This choice of words is not casual, but rather is inspired by my visit last year, while preparing an article for The Fretboard Journal, to the shop of renowned luthier Dana Bourgeois. He is among the handful of luthiers who spurred the return, to the steel-string guitar industry, of the voicing techniques once popular in factories before the 1940s and '50s: techniques that were abandoned in an effort to streamline production and to produce guitars that were more durable.

Bourgeois begins by considering the manner in which the instrument will be used—the player's touch and picking style, the gauge of strings that he or she prefers, the desired degree of loudness, and so forth—and selects for the top a pair of spruce boards of the precise degree of required stiffness. Bourgeois and his co-workers then brace the top; tap it at various different nodal points, listening for a particular tone; slightly thin the braces; then re-tap, re-listen, and re-thin until the desired tones are achieved. After the top is attached to the body but before its binding is attached, the luthier flexes the top and, if the desired flexibility is not observed, he or she gradually thins its periphery. Finally, after the top has been trimmed, a luthier trained in the procedure taps the bridge to ensure that the top is pushing back to just the right extent.

This method of matching the instrument—which is, of course, an acoustical source, amplifier, and loudspeaker all in one—to the player is so natural, so reliably right, that one wonders why it should be done any other way.

At roughly the time when the larger guitar companies abandoned the notion of voicing their instruments, the leaders of the domestic audio industry decided that, in their quest for flatter frequency response, greater frequency extension, and more "precise" stereo imaging, they would be better off designing their loudspeakers to be unresponsive—that is, to perform less efficiently at transforming electricity into movement, which is the single most important thing a loudspeaker does.

How then, you might wonder, would the consumer drive such an unresponsive loudspeaker? By buying a much more powerful amplifier, of course—because, thanks to Our Friend the Transistor, power is cheap. Besides, all competently designed amps sound the same. Right?

Parallel
Last October, on the day before the last leaves fell from the trees behind my house, a pair of Shindo Laboratory's newest mono amplifier, the D'Yquem ($24,995/pair), arrived at my house for a brief visit (footnote 1). The D'Yquem is named for Chateau d'Yquem, which produces the most expensive and universally well-regarded of sauternes. Novelist Thomas Harris had his star antagonist, Hannibal Lecter, buy a 1961 d'Yquem as a birthday present for protagonist Clarice Starling (it was never delivered), and Julia Child, Ernest Hemingway, and Thomas Jefferson were among the wine's most notable admirers.

Each D'Yquem amplifier produces up to 18W from a parallel pair of 300B directly heated triode tubes, operated in single-ended mode (and thus in pure class-A). Shindo Laboratory claims for the amp a signal/noise ratio of 110dB, an input impedance of 120k ohms, and harmonic distortion of 0.01%. Although an output-impedance specification is not offered, I assume that the D'Yquem's single secondary Lundahl output transformer is, in typical Shindo fashion, optimized for loads of 8 ohms and (especially) higher.

A glimpse inside reveals some interesting variations on the design and construction details I've seen in Shindo's other amplifiers during the eight years I've followed the brand. Like most amps in Shindo's current lineup, the D'Yquem doesn't use tube rectification, and while every semiconductor-rectified Shindo amp has an internally mounted EY88 diode tube between the AC transformer's high-voltage secondaries and ground—it slowly ramps up the rail voltage to help prolong the life of other parts—the D'Yquem has two EY88s, wired in parallel. Strange! Another unusual doubling-up is seen in the bias supply of this fixed-bias amplifier, where the two output triodes are served by no fewer than four potentiometers. The 300Bs are themselves a different sort for designer Ken Shindo: contemporary Russian-made tubes bearing the oft-traded Genelex label, instead of the Western Electric or Cetron tubes I've seen in all of Shindo's other 300B models.

As he has with all of his amps and preamps, Ken Shindo voiced the D'Yquem with a mixture of vintage parts from his reportedly vast stock of same—Sprague Vitamin Q and Black Beauty signal capacitors, a lovely old Mallory electrolytic cap for the bias supply, NOS Philips 6AW8A dual-triode/pentode tubes for input gain and buffering—and other parts that are decidedly modern. The latter include a smattering of French polypropylene-film capacitors from Solen, an uncharacteristically large (for Shindo) dry-electrolytic reservoir cap for the main power supply, a new type of Japanese ceramic-substrate resistor for select applications in the signal path, and the Swedish Lundahl transformers that Shindo has come to prefer for most of his products.

The DiaLogue tube amplifiers from PrimaLuna have, at their heart, a clever idea of their own: an output circuit that is user-switchable between triode operation, in which the screen grid of a tetrode or pentode power tube is defeated by means of connection to the tube's anode; and Ultralinear operation, in which the screen grid of a tetrode or pentode carries a portion of the AC music signal, supplied by a tap on the output-transformer primary, in a feedback-like effort to reduce distortion and increase power. Fans of the former often report a sweeter, more tubey sound, while fans of the latter report a tighter, more detailed, more timbrally neutral sound. Audio enthusiasts are given to reporting any number of things.

Here's the difference between PrimaLuna's clever invention and all of those self-centering turntable platters and "sonic hologram" generators: None of the latter really caught on in the long run, their makers having moved on to ever more newsworthy gewgaws, while PrimaLuna's Triode/Ultralinear circuitry has endured since 2006—and has apparently succeeded in the marketplace. (The company itself got started in 2000.) It has also spread to even more products in the PrimaLuna line, the most recent being the DiaLogue Premium amplifier ($3199), which offers 25Wpc in triode mode, 42Wpc in Ultralinear.

Description
The PrimaLuna DiaLogue Premium has more than one clever trick up its sleeve. Flip the rocker switch on the right side of its chassis and this erstwhile EL34 amplifier can now accept power tubes from the KT88/KT120 family of pentodes. Flip a switch on the rear panel and this stereo amplifier is now strapped for use as a higher-powered monoblock, after which you have a choice of buying one more DiaLogue Premium or discarding one of your loudspeakers. (You know which route I'd go.) Sit back and ignore both of those switches and the DiaLogue Premium still offers: an LED-based Bad Tube Indicator system for the output tubes; a Power Transformer Protection circuit; an Output Transformer Protection circuit; and an AC Offset Killer circuit intended to eliminate hum and promote more silent spaces between the notes. Choices!

Enduring in the DiaLogue Premium amplifier is an updated version of a design innovation that has, from day one, characterized all of PrimaLuna's amplifiers: their Adaptive AutoBias circuit. In contrast with the traditional application of the term, whereby an auto-bias circuit is one in which the signal grid automatically takes on a negative bias relative to the cathode simply by using a resistor to raise the latter above ground, PrimaLuna's Adaptive AutoBias keeps the cathode at or near ground, and applies to the signal grid a negative voltage that is subject to continual change, in response to amplifier demands and the aging of the power tubes (footnote 1).

Back to the Triode/Ultralinear choice: In the last PrimaLuna amp I reviewed, the DiaLogue Seven monoblock (see Stereophile, January 2010), chief designer Herman van den Dungen and his engineering team effected that switching capability by putting two separate output transformers into each mono amplifier. Yet in a stereo amplifier such as the DiaLogue Premium, a total of five large transformers—two Ultralinear output transformers, two triode output transformers, and one sizable power transformer—would add to the product unwanted weight, bulk, and electromagnetic fields. Thus the DiaLogue Premium has only a single stereo pair of output transformers, the primary sides of which are bordered by small circuit boards containing relays and transistors, presumably to reconfigure, on the fly, the various primary taps.

Indeed, small circuit boards stuffed with relays and transistors are the order of the day here—the AC Offset Killer board alone has a full dozen transistors—yet these are not in the signal path, which itself is executed with conspicuously neat point-to-point wiring and such high-quality parts as Solen capacitors, Takman resistors, and robust ceramic terminal strips. The DiaLogue Premium has three 12AU7 small-signal tubes per channel, including voltage-gain stages and a long-tailed pair for phase inversion. The power supply uses silicon devices for rectification—individual diodes for the rail, a full-wave bridge rectifier for the heaters—followed by a pi filter incorporating two large Nichicon reservoir caps and a neat proprietary choke. As I've found is typical of PrimaLuna, whose products are designed in the Netherlands and assembled in China, the build quality of the DiaLogue Premium's circuitry and casework is excellent.

Installation and setup
One could easily go overboard writing about the PrimaLuna DiaLogue Premium: Here's how it sounded with EL34s in triode mode, and the same tube in Ultralinear mode, and the KT88 in triode, and . . .

Because the amp is supplied with EL34 power tetrodes as standard, that's where I started. And because the review sample, as shipped, was set to Ultralinear mode, and because the remote handset, which controls only that function, arrived with a dead 2032 3V battery, and because that function can't be controlled from the amp itself . . . well, that, too, is where I started.

I enjoyed the detailed and pleasantly punchy sound of the Premium in EL34 Ultralinear mode. But when I refreshed the handset's battery and restored the Premium's switchability, I was even more impressed with the amplifier's Triode-mode performance, which, through my very sensitive Altec Valencia speakers, had a more pleasing and more "organic" bottom-to-top balance of performance, while appearing to sacrifice nothing in terms of detail or temporal accuracy.

And then, when I replaced the Premium's stock, PrimaLuna-branded EL34 tubes with the Russian-made Tung-Sol KT120s sent to me, for trial, by PrimaLuna distributor Kevin Deal, I was even more impressed with those—an $80 option at time of purchase. I also preferred to use the KT120s in Triode mode.

To confirm those impressions, I did a considerable amount of switching, back and forth, of tubes and switches. And there came a day when I discovered that I'd spent a few hours listening with KT120 power pentodes, but with the tube-selector switch—which controls bias voltage—set for EL34. After making the correction, I noticed that not only was the switch's effect subtle, but that I actually enjoyed the sound of the KT120 tubes set for EL34 bias.

Curious, I opened up the DiaLogue Premium and made a few cursory measurements, at which time I discovered that the position of the selector switch makes a difference of less than 2V for the same tube: –54.6V when set for KT120s vs–56.4V when set for EL34s. (I suppose one might see more or less of a voltage difference when changes in switch position are accompanied by changes in power tubes, given that different tubes can draw different amounts of current.)

Be that as it may: The reader should note that all of the listening observations that follow were based on the DiaLogue Premium used with KT120 power tubes operating in Triode mode, with the tube-selector switch set for KT88/KT120.