What if one wanted transistors to sound like hollow state? In many ways, it’s an entirely Quixotic proposition. No matter what, transistors will never be tubes. Or vice versa. It’s Rudyard Kipling’s never the twain shall meet. Transistors will never have swarms of electrons zip through a vacuum. Whether, Bob Carver like, one models their transfer function or ends up with equivalent output impedance; uses output transformers or McIntosh-like autoformers… treating two inherently different output devices in the same way, eve arriving at similar THD via other chicanery can’t offset that these parts will always be fundamentally dissimilar. Many of the Nelson Pass FirstWatt amps in their topologies share things with basic tube-amp circuits but they neither sound like them nor were they intended or claimed to do so. To end up with a badly hobbling sentence: Similar isn’t the same as same.
What if one rebiased the question? What if an experienced tube-amp designer was tasked—or tasked himself—with designing a sand amp? How would he approach it? In the case of one Alexey Syomin from Russia, extensively chronicled in these pages already, his resultant class A/B White Knight integrated does get massive output transformers. It even uses a 5Hz-300kHz stepup transformer for passive 32V gain. Transformer gain we know of from low-output cartridges. It’s rarer in power amplifiers though FirstWatt and Audio Consulting are two easy precedents. In typical SET fashion, the White Knight uses no negative feedback. But then, neither do solid-state electronics from Ayre Acoustics or others. For balanced power, the overbuilt bipolar supply runs on four vertically aligned toroidal line transformers. This array pursues maximal noise rejection on the AC mains. It also contributes heftily to the final weight.
“Each channel has two power transformers, one for ‘+’, one for ‘-‘. To wind these, we use high-precision machines. They guarantee perfect symmetry of the power supply poles. Transistors suffer distortion when rising temps lead to higher current conduction and eventual failure. My aim was to create a constant quiescent current regardless of temperature. Other parts require their own thermal stabilization to operate best. In our case, transistors operate at 70°C, capacitors at 60C° and resistors between 40-50C°. We stabilize our transistors with Beryllium oxide couplers. Our output module runs 24 powerful transistors, 6 for thermal control, 18 for current gain. Capacitors too alter behaviour with rising temps. We stabilize this with different thermally conductive compounds. Partitions inside our module prevent these compounds from mixing. This locks in optimal working parameters for different parts. The only downside is extreme complexity for assembly of our potted modules. It’s time consuming and uses 6kg of exotic compounds to add serious labour and parts costs.” Military-grade sealed relays act as input switchers. ‘Direct’ mode bypasses the volume control for the first input. It converts our integrated into a pure stereo power amp. Delivering 90/180wpc into 8/4Ω, the first 20 watts operate in class A. The attenuator is a simple variable resistor at the amplifier input.
Whilst one wouldn’t expect the White Knight to sound like a traditional valve amp, it was designed by a classic valve-amp expert. It’s thus accompanied by very specific sonic expectations; runs on ‘heavy irons’ like a premium tube amp; and combines a passive stage for voltage gain with a single current buffer where transistors are exploited solely for what they do best: amplify current. This mirrors SET lore with a minimum-stage architecture. Diverging from common SET specs meanwhile are 10-250’000Hz bandwidth ±0/3dB, a damping factor of 950, claimed SN/R of -140dB and THD of below 0.001%. Given these figures, one knows going in that the White Knight doesn’t really behave like a tube amp. Knowing who designed it, one simply wonders what, exactly, itdoes sound like.
When Alexey emailed that he’d delivered the amp to Russian customs in Moscow for export, a smiley face added that “I hope they don’t break anything”. Given what looked to be a Blackbird-type chassis loaded with heavy iron, I seconded his thoughts. Two prior S.A.Lab deliveries had incurred shipping damage; one with a broken transformer bolt to suffer a free radical donut; one with broken stand-offs to net a very precarious heavily loaded PCB. “After quite the experience trying to send you the White Knight amp, I finally succeeded. I don’t know why but customs requested a fumigation treatment. We had to unpack it from the original box (customs said it looked too military, hence suspicious) and repack it in two other boxes so total ship weight decreased from 68 to 56kg.” So much for smooth governmental cooperation with small domestic entrepreneurs trying to do the export business.
The sonic map. From the middle path of flat-lined neutrality to subtle deviations of strategic voicing to more specialized/excessive attempts… the assembly of hifi systems in wildly different rooms offers its owners quite the palette to paint with. Regardless of personal bias and goals, it’s pretty much undisputed that—disregarding room acoustics which most renters and buyers are stuck with—the interface of amp to speaker is most critical. Many speakers particularly of the vast ported majority benefit from higher damping. The lower the amp’s output impedance, the stronger this effect (within reason since the resistance of speaker cable, contacts and voice coil quickly cuts into 4’000+ über figures which look good on paper but soon stop to matter). Speakers with very light-coned drivers and ultra-potent magnets expressed as steep sensitivity ace self damping. They can act dry, brittle and quite bass shy when (over)powered by such amplifiers. It’s one reason why so often, 100dB widebanders sound better with valve amps of apparently inferior—far lower—damping aka high output impedance.*
This subject of damping extends to musical gestalt. Sitting very close to stage nets a fast energetic very direct highly dynamic sound. This suffers little to no alteration from venue acoustics. The farther away one sits, the more venue acoustics dominate to introduce reflective reverb aka hall sound. This nonlinear skewing adds fuzz and softness by lengthening decays. It creates connective tissue as new stuff between notes. Sounds overlay, blend, intermix and cast half shadows. Timbres get richer but separation and dynamics diminish. So do transient heat and timing accuracy. That’s why marching bands don’t belong in a church. If you think on it, this makes for three dimensions where more or less damping figures. First we have recorded ambiance. This could range from very dry to very wet. It includes actual venue sound, artificial reverb and microphone placement effects. Second comes the playback venue. That’s the sound of our own room. Third there’s the interaction of our hifi hardware. That too moves a slider on the “more leading edge or more trailing fades” axis. Obviously #2 and #3 have nothing whatever to do with the recording. They’re side effects of reality. Only idealists talk of pure windows on their recording. For the rest of us, it’s a meaningless abstraction. Even the purists cannot eliminate the sound of their room and system as two big filters andchangers on what was recorded.
Once we drop the chastity belt of unachievable ideals, playback becomes co-creative. Our choices on gear/setup are a self expression to generate the type experience we find most compelling. There’s no right or wrong. Playback, sitting behind the studio mixing console and live music are three very different states. The first is ours to shape as we please. If we favour a central front-row musical perspective, we’ll want more damped dry acoustics in our playback room; perhaps sit closer to the speakers and toe them in sharply to minimize room interactions; and probably fancy high damping factors and electronics with maximal speed and clarity. If we favour a front-row balcony seat instead as more of a bird’s eye view on the musical action, we’ll want more lively reflective acoustics; perhaps sit farther away, with speakers pointing straight ahead; and perhaps appreciate lower damping factors and electronics which prioritize decays and tone. It shouldn’t surprise that classical listeners with regular concert attendance often favour that. It’s closer to their live reference. Talking in generalities, the first listener will often prefer solid state whilst the second ends up with tubes somewhere in the signal path. On specs, the White Knight belongs to the first group. On gestation and design genetics, one expects influence from the second camp. Where would it move our slider?
Delivery of a wooden crate inside a bigger white-painted wooden crate separated by beaucoup styro pellets was uneventful. The White Knight emerged undamaged from its liner. With another assignment in the final stages in the big system, it first went into our entertainment room. Here a pair of Kroma Audio Julieta super monitors were beginning to acclimate their internal wooden parts to our far mellower lusher milieu over Granada’s hot and dry summer where they’d come from.
In this particular hardware context, insertion of the White Knight caused a strong ground loop where with the Job 225 amp, there’d been none. My lazy fallback for the few days it’d be here was to run it off the Esoteric C03 preamp not direct but line-level, opening its own volume control to just 12:00. This attenuated the hum to near inaudibility. If I had a reoccurrence in the main system, I’d have to systematically lift various grounds by figuring out wherever I had stored away the cheater plugs after our recent rather massive move to Ireland. For break-in purposes however, this setup was just fine to put some 50 hours of calisthenics on the speakers and amp alike.
An immediate observation to make was that replacing the Job 225 clearly stepped down quicksilvery lucidity and speed and upped tonal weight and a slightly more languorous mien instead. Given that the cabling here was solid-core pure silver Ocellia inside a crushed crystal sleeve; and that the synthetic stone speakers reduce typical box-talk fuzz to very audible degrees… this shift on the attack/decay axis into the second half was attractive and without an undue loss in apparent resolution though step down it did.
Once in the big system, the ground loop vanished as expected for proverbial grave-like silence. Not even the on-tweeter ear could make out any operational self noise. In this space, the White Knight would compete against the $6’500 Pass Labs XA30.9 and $3’000 FirstWatt F7.
Barking up the wrong tree? Lady Circumstance wrapped the Lio DHT review just before the White Knight got busy. This conclusively settled an old but basic argument. At least in my experience to date, the only way to make transistor amps sound like proper tubes is with direct-heated triodes in the preamp stage. But not just any. Vinnie Rossi’s is a spud circuit. This means no driver tubes, no output iron. It does mean 500kHz@-3dB bandwidth with stock electro-harmonix Gold Grid 2A3 bottles; and zero noise! This rare affair potentizes the languid, fluffy, spatially huge triode aroma well beyond personal ownership takes on Art Audio’s PX25, Yamamoto’s 45 and 300B amps and Woo Audio’s 300B monos. The only real exception to that lot had been my friend’s Berning Siegfried. That sounded fundamentally different.
The Lio trick is still other. Any proper transistor amp will transcend typical SET limitations on bandwidth, driver control hence bass grip, distortion density and SN/R. What the DHT spud + transistor amp combo does is distill, concentrate then dominate with what’s best about direct-heated triodes. This eliminates all of their weaknesses. So it is still not the same. It’s better. It even works with solid-state amps of Job 225 or Crayon CFA-1.2 stripes as just two of their kind I tried [Crayon setup below]. Obviously truly ideal SET/speaker mates can downplay or perhaps even entirely overcome typical SET limits. It simply requires very specialized transducers. This seriously narrows your window of choices. Meanwhile Vinnie’s spudster pulls off that stunt with arch-normal low-efficiency speakers using ceramic drivers. Tube wranglers, put that on your breakfast menu!
All this by way of saying that the White Knight couldn’t duplicate that effect even with our tubed Fore Audio DAISy1 DAC and tubed Nagra Jazz preamp in the loop. 12AU7/6922-type small triodes won’t give off that special DHT aroma no matter how much you might like them to. Given Lio’s powerful proof to the contrary, Alexey’s amp really had to be judged as a transistor amp, never mind our two-page back story. In the sweat spot, none of that mattered. The White Knight sounded like a transistor amp. Battling our XA30.8 obviously set very high standards. If the S.A.Lab ended up not being their full equal, it should actually be expected, never mind its 30% higher tariff. Nelson Pass has plied his specialized craft for more than three decades. Today he’s called a legend in it. Alexey’s core expertise is valve circuits. He hasn’t yet put in the same transistor time. Regardless of advertised quality, could his output transformers really be better than none?
TIme to progress from back to front story. To reshuffle the deck accordingly, the Lio DHT was benched in favour of the D1 DAC/pre from COS Engineering. Though a fully balanced deck, it would run both Pass and S.A.Lab single-ended to use the same cable into the latter’s direct-mode input. This signal path was pure solid-state.
Pass Labs XA30.8 vs. White Knight. The core difference here was about a pervasive softness associated with focus and grip. The Russian played it more polite and gentle/genteel. This was a full-bandwidth effect but perhaps most dominant in the bass where the American exerted more apparent woofer control for more profundity and foundation anchoring. Whilst not affecting subjective soundstage perspective, the White Knight sounded more distant and farfield because of its minor vagueness. The XA30.8 sounded more powerful with the usual things this entails: more propulsion, drive and forward communicativeness; more firmness, substance and slam when called for. This offset still enlarges when COS D1 and Pass connect via XLR (which merely is an observation about how they interact, not about either sounding better in RCA or XLR mode).
Tonal balance was very comparable. This parked the White Knight in the same general vicinity of minor darkness and warmth which distinguishes the Pass from our resident Crayon and Job ‘speed’ amps. If you insist to assign this twin constellation of qualities—the second being a more distanced attack behaviour—to tube sound, the White Knight certainly belongs. It’s simply not exclusive to valves nor true for premium DHT implementations. To reconnect with damping and regardless of its output impedance, the S.A.Lab sounded less damped than the grippier Pass and in fact, slightly underdamped. Whether or not this was due to the output irons or stepup transformer I couldn’t say. I will say that prior experiments with transformers for voltage gain did share some of this more limpid de-stressed gestalt which bleeds out a certain toned tension.
That can be a key ingredient for the row 15 symphonic perspective in a very large hall. It’s not for a close-to-stage club performance or most hard-hitting modern music for that matter. From this one suspects that Alexey used classical music and its general perspective as the overriding design reference also for his transistor amp.
FirstWatt F7 vs. White Knight. Now the offset in coin grew to 60%. One could acquire three F7 for one white Russian. This juxtaposition now closely matched class A bias—a pure 25 watts for the FirstWatt, the first 20 watts of the S.A.Lab—whilst piling on headroom for the class A/B challenger. For that one expects a broader slab of potential speaker mates. The F7 should eliminate certain beastlier loads. Would any of that telegraph with our easier speakers?
Only for operative voltage gain. Here the Russian’s was a lot higher. That meant more headroom. Sonically the presentation was essentially on par except for the bass. Here I found the F7’s to be texturally more seamlessly integrated. Aside from that—and whilst applying far higher parts density, power supply complexity and decidedly more output devices—the White Knight in this context behaved rather similar to the F7 if not with its innate relaxation where the FirstWatt was tauter. It then added an integrated’s functionality though no remote, a big oversight at its price. For sonics we recall the F7’s award. Pulling up alongside it really was a compliment to Alexey Syomin even if his way of getting there wasn’t as elegantly simple and cost-effective as that of the legendary Nelson Pass. With X marks the spot established, it was time to reshuffle the speakers for something that’d play maximally to the White Knight’s fingerprint. I wanted no fainting damsel in distress. I envisioned more of a butch biker babe. Enter Zu Audio’s Druid V. With above average sensitivity and a pro/vintage-style hard-hung suspension for its 10.3″ widebander augmented by a 15lbs 2cm exit compression tweeter of top pedigree, it doesn’t appreciate excessive amplifier damping either electrically or texturally. But it arrives with a sporty feisty attitude as part of its voicing and behaviour. It’s dialled to do justice to modern music.
To ratchet up that particular druidic aspect required removal of the spike shoes which I’d inserted previously for the speedier Crayon amp. That removal lowered the speakers’ floor gap. Less gap equals higher damping on the widebander if also slightly less bass extension. To fully optimize, I then inserted the Zu Event cables to match what the speakers are wired with internally. We don’t own bright strident forward boxes whose otherwise unattractive behaviour would most likely have doubled as grease cutter and accelerator. Personal taste really desired more gumption and kick. Here the Druid V’s far from sharp but instead vigorous punch really was my best/favourite speaker option. Whilst it didn’t rewrite the White Knight’s signature, it wrote it out a little smaller. I now was on the fringes of the gentrified ‘hood, not in its centre. And that really describes the White Knight. Its innate softness and temperamental chill factor dominated. Whilst perhaps related to a certain tube idea many entertain, it lacked that genre’s freely breathing gush. The White Knight was temperamentally buttoned up. It’s what had me turn over a number of pages in my search for ideal partners. If this was general tube sound, it also was exactly what got me out of tube amps.
It’s a mighty fine balance. Leisure Club Med mode is brilliant when toned, not if allowed to go to seed. With the Druid V, muscle tone was back. For me and how real emotional conviction talks, the presentation simply remained a bit too laid back. But personal preference isn’t our focus. The elemental thing is that the White Knight’s calm body temperature was fixed. It was invariable and built in. I could play to or against it but never around or through it. I strongly suspect its miles of signal-path transformer wiring. No matter one’s hardware choices, those obviously don’t unwind. They seem to be the determinant factor for better or worse. Given my DHT experiments as presently culminated in the Vinnie Rossi LIO, I happen to think that shifting responsibility for the floating spacious flavour onto such a preamp stage without transformers, then letting transistor like the XA30.8 excel at their strengths is more effective. Those who disagree or want it all in one box without actual tubes… they could be tailor-made to investigate today’s very civilized and calm White Russian.
PS. Post publication, I received this from designer Alexey Syomin: “I have some news about the White Knight’s future pricing. I’ve modified the technology which helps to significantly decrease the retail price. It should be about €1’600 in Russia and I assume no more than €1’900 in Europe; quite a drop from €9’000. This will be effective by October. The main technological changes are no use of thermal compounds and Mosfets for output transistors.”