This review page is supported in part by the sponsor whose ad is displayed above

Reviewer: Srajan Ebaen
Source: Zanden Audio Model 2000P/5000S; Opera Audio Reference 2.2 Linear; Raysonic CD128 [on extended loan]
Preamp/Integrated: ModWright SWL 9.0SE; Music First Audio Passive Magnetic; Bel Canto Design PRe3; Wyetech Labs Jade; Melody Hifi I2A3 [on extended loan]; Supratek Cabernet Dual [on loan from owner]

EQ: Rane PEQ55 active merely below 40Hz
Amp: 2 x Audiosector Patek SE; Yamamoto A-08S; FirstWatt F3 & F1; Bel Canto e.One S300; Eastern Electric M-520; Yamamoto HA-02
Headphones: AKG K-1000 w. hardwired Stefan AudioArt harness; audio-technica W-1000
Speakers: Zu Cable Definition Pro in custom lacquer; Anthony Gallo Acoustics Ref 3.1; Mark & Daniel Ruby with Omni Harmonizer

Cables: Zanden Audio proprietary I²S cable, Zu Cable Varial, Gede, Libtech and Ibis; Stealth Audio Cable Indra, MetaCarbon & NanoFiber [on loan]; SilverFi interconnects; Crystal Cable Reference power cords; Crystal Cable Ultra complete wiring loom; double cryo'd Acrolink with Furutech UK plug between wall and transformer
Stands: 1 x Grand Prix Audio Monaco Modular five-tier
Powerline conditioning: 2 x Walker Audio Velocitor S fed from custom AudioSector 1.5KV Plitron step-down transformer with balanced power output option
Sundry accessories: GPA Formula Carbon/Kevlar shelf for transport; GPA Apex footers underneath stand, DAC and amp; Walker Audio Extreme SST on all connections; Walker Audio Vivid CD cleaner; Walker Audio Reference HDLs; Furutech RD-2 CD demagnetizer
Room size: 16' w x 21' d x 9' h in short-wall setup, with openly adjoining 15' x 35' living room

Review Component Retail: $8,900/pr for NVS-003Gs | $4,900 for SE-MkII remote-controlled preamp

Manufactured with pride in Ho Chi Minh City, Viet Nam. Thus begins our journey today. Surprised? Now read the ingredients. Cardas gold/Rhodium RCA input jack. Audio Note Rhodium binding posts. Riken resistors. Auricaps. Jensen paper-in-oil capacitor. Mundorf power capacitor. Nichicon filter capacitor. WBT solder. Tango output transformers. Ceramic tube sockets. Barian Kingwood casings. On second thought, let's strike the fine Japanese Tango iron, shall we? It does feature in Navison's other valve amps, just not in the NVS-003G monos. As OTLs, their quartet of triple-nipple 6C33C-B triodes couple to your speakers' crossovers without an output transformer for 75 watts of Class AB push/pull power. Better yet, if you run active networks ahead of the amps or no crossovers at all, these  Russian bottles will couple truly direct to your driver voice coils.

This parts list reads like something off a modifier's option menu. It's the exotic stuff which stock machines often get upgraded to after plenty of extra cashish -- for American or EU specialist labor in small hotrod shops -- has exchanged hands first. Not here. It's all part of the standard trim package. Navison Audio was founded in 2000 for private research by a group of Vietnamese engineers with many years of tube experience. These folks are very serious about quality. They're rightfully proud about what goes into their creations. Their 61lbs OTL monos are very impressive specimens of the art. 27 inches deep, 15 inches wide and 9 inches tall, they'll take pride of place whether you want them to or not. Input sensitivity for 21dB of gain is a wisely low 2.5V for full output. A preceding preamp will actually get to work out a bit. It won't be completely stuck in first gear as happens so often with ultra high-gain amps whose input sensitivity could sit well below 0.5V. With modern sources, that tends to park any partnering preamp a few levels below the ground floor of unity gain. The preamp gain is thrown out the window yet its self noise gets magnified manifold by the amplifier. Not here.

The 003-Gs run one 6SN7 in the input and two in the driver stage. Hum and noise are given as minus 75dB below full power. Output impedance isn't specified other than 8-ohm speaker connection, recommended. If you're familiar with the general concept of output transformer-less valve amps, you already know that tubes are high-voltage low-current devices yet speakers want to see high currents at relatively low voltages. The insertion of the impedance-matching step-down transformer in conventional tube amp outputs bridges that gap and blocks DC. Without one, impedance matching has to be accomplished by the tubes themselves; by paralleling many of them; by adding cathode followers; and with sufficient negative feedback. Lacking the transformer, speaker protection requires other means, say fuses or an output capacitor.

As Bruce Rosenblitz of Transcendent Sound tells us in his primer, a 300B SET with 700 ohms of plate resistance gives us 700 above the line of the left formula on how to calculate amplifier output impedance. If its transformer primary were 3000 ohms divided by an 8-ohm secondary, we'd get 376 below the line. Output impedance
then becomes 700 divided by 376 = 2.13 ohms. Without the transformer, it'd be a ghastly and completely useless 700 ohms. If we paralleled ten 300Bs just for argument's sake (and because good ones are so affordable - right), we'd still net an unacceptably high 70 ohms without a step-down output transformer. Plainly, 300Bs -- and many other tubes -- simply aren't practical for such applications (though that hasn't prevented "the world's only 300B OTL"). According to the Tube Lovers Anonymous paper, the very first commercial OTL only worked into a special 500-ohm loudspeaker. Needless to say, it did not become a commercial success. "With the introduction of the 6AS7G twin-power triode in 1946... plate resistance was listed at 280 ohms per section. If enough tubes could be combined in parallel, it wouldn't be too difficult to drive a typical loudspeaker load, which at that time was around 16 ohms." Later cathode followers were added to drive down output impedance further. Futterman's contribution in 1954 was to propose "that the cathode resistor of a split-load phase inverter be returned to ground through the load, rather than to ground, as was the normal practice... If four 6C33C-B tubes were used in the Futterman OTL, biased at 145V and 400mA each, the open-loop output impedance would be 25 ohms - coincidentally, about the same as the push-pull cathode follower version."

A later variation on the Futterman circuit with the same four 6C33C-Bs dropped "open-loop output impedance to about 6.8 ohms, 4 times lower than the original Futterman!" The original exclamation mark of our verbatim TLA quote [italic emphasis ours - Ed.] reminds us. Impedance matching without a transformer is serious business for OTLs to become viable. 1955 saw Henry Wiggins' Electro-Voice Circlotron circuit, a dual single-ended floating bridge affair. Cecil Hall was awarded a US patent for it around the same time. Another accounting lists Tapio Köykkä instead as having held the very first two patents for Circlotron architectures, albeit limited to his native Finland. His schematic published in the German radio magazine Funktechnik (issue 7 of 1953) to predate the Electro-Voice product. Then Jim Bongiorno used a Circlotron variation in his Great American Sound transistor amps during the 70s.

But it wasn't until the late 1980s that the first commercial thermionic Circlotron iteration manifested as Ralph Karsten's groundbreaking Atma-Sphere MA-1. Still according to the same paper, its open-loop output impedance was about 10.6 ohms with our reference quartet of 6C33C-Bs, i.e. slightly higher than the preceding Futterman variation but now fully symmetrical and balanced (Futterman's original circuit suffered asymmetries and instability, i.e. it was prone to oscillation). Atma-Sphere's implementation allowed the elimination of negative feedback altogether.

Still, a lowish damping factor remained. It benefitted from additional impedance matching with an externally inserted tapped autoformer (the so-called Speltz Zero boxes above, championed by Karsten) to become truly copasetic into lower-impedance speakers. Naturally, purists regard the addition of any outboard iron as utterly contrarious to the entire OTL endeavor. Hence speaker matching -- particularly their impedance behavior across the bandwidth -- remains vital if you want to optimally operate an OTL as an OTL rather than cheat with an add-on transformer. (Incidentally, impedance multiplying with an autoformer not only raises the nominal impedance. Impedance swings will be multiplied by the same factor. If your 4-ohm speaker swung between 3 and 12 ohms without the autoformer, raising its nominal impedance to 16 ohms means it'll now peak at 48.) That said, adding just 10dB of feedback to an output transformer-less Circlotron circuit can lower its output impedance to below 1 ohm into a 4-ohm speaker. That's better than many transformer-coupled SETs. So much for memory lane. You'll appreciate why, just for starters, we do want to know what the output impedance of the 003-G monos is, exactly. At full power with 4 output tubes on deck, it is 0.9 ohms. At half power with 2 output tubes, it is 1.6 ohms.

Incidentally, the paralleling of valves for impedance-lowering purposes is done in preamplifiers as well. It is exploited by Conrad Johnson's composite triode in the ART preamp as well as Navison's own SE-MkI and II. Both Vietnamese preamps use eight 6H1-EB triodes, a relative of the 6922. Only with headphone impedances running as high as 600 ohms are OTLs with just one output tube per side feasible.

From our general overview of the genre, we can come to certain conclusions. 1/ Since plate resistance lowers by applying higher plate voltages, OTL designers will often be compelled to run their tubes hot. This will create physical heat and shorten tube life. 2/ To drive low-impedance speakers, massive paralleling of tubes is required to make for an acceptably low output impedance though the obvious increase in power can disguise this primary goal. More tubes equal yet more heat. 3/ If a tube arcs, there's no transformer to shield the speaker. Various fused or other protection schemes become vital to prevent very high voltages on the amplifier outputs. 4/ The same high voltages develop within the circuit. Prying fingers should know what they're doing before opening the bottom cover of a recently shut-down OTL amp. 5/ In massively paralleled powerful amps, one could pull out tubes without hurting operation depending on how tube biasing is accomplished. This would merely lower output power and diminish drive since output impedance will invariably increase.

Questions to ask any given OTL amp designer are thus - how much thermal impact will running a given amp have on the room; what kind of protection circuits against catastrophic tube failure are provided; what is the operative output impedance to assist us in determining copasetic speakers. The latter is doubly important since another peculiarity of this amplifier breed is its
inverse power response. It increases with ascending rather than descending impedances. Think Krell but upside down or counter clockwise. Power transfer into 16 ohms will be superior than into 8 ohms, i.e. higher. Into 4 ohms however, it declines well below the 8-ohm power rating. The most copasetic speakers for OTLs thus tend to be 8 ohms minimum. Ideally, they'd be 16-ohm or higher still. A well-matched speaker on a smaller OTL -- i.e. one with fewer output tubes -- shouldn't drop below 7 ohms in the bass or midbass where power transfer becomes critical and patently audible when insufficient, as under-damped and loose ill-defined bass. (By way of examples, consider the Atma-Sphere range of OTL amps. The 60-watt monos utilize 6 power tubes for 4.1-ohm output impedance, the 140-watt monos 14 tubes for 2.3 ohms and the 220-watt monos 20 tubes for 1.75 ohms.) When speakers and amp are properly matched, aficionados of the OTL breed will tell you that you'll never miss any of "those phase shift inducing, induction leaking, current robbing, amplitude attenuating mega yards of wiring inside the usual output transformers". Instead, you're to enjoy a new-found purity, speed and directness which transformer-coupled amps simply aren't capable of. End of OTL sign-up spiel.

For a counterpoint, I have also heard OTLs sound lean and harmonically bleached. Likely this was because of sub-optimal speaker matching and the concomitant audible lack of drive. Regardless, it wasn't remotely as magical as had been promised. Once again it seems best to simply leave the salubrious spin work to politicians. We'll focus on the subjective implementation of any given technology, i.e. what it sounds like to the ear in the listening seat. Based purely on the general technical background already covered, it's clear that OTLs by definition are far from the magic bullet for conventional low-impedance loudspeakers which so completely dominate the market today. This simple fact is further validated by the sheer scarcity of manufacturers selling OTL amplifiers. Non-devotees may have a hard time coming up with more than four makers. To name all of Absoluta, Atma-Sphere, Berning, Croft, Einstein, Graaf, Joule Electra, Naked Truth, S.A.P., Silvaweld and Transcendent Sound will take a bit of digging for most of us not already inside the OTL tent.
Craig Uthus' Zana Deux OTL for EddieCurrent gets away with a single 6C33C-B per side to drive 30 to 6000-ohm headphones

TenorAudio and Fourier are defunct by now and BAT's popular tube amp uses a transformer-coupled variant of the Circlotron circuit. (Some slide rulers even argue that David Berning's RF conversion transformers defy the classic OTL definition though his revolutionary circuit design brilliance is never questioned.)

With such limited commercial choices, I was thrilled when today's Navison Audio opportunity was presented. Surveying their short 5-model line, I deliberately opted for the 003-G OTLs over their 572 or 211 transformer-coupled SET brethren. Besides using these mono blocks to profile an exciting company that's new to the West, I'd educate myself on an established amplifier topology I'd never yet auditioned in a familiar context before. To answer the questions I mentioned earlier as relevant for any OTL, the NVS-003Gs are both fuse-protected and capacitor-coupled at the output. Like all Navison amps, they're fixed bias. The OTLs run at 110mA, the NSV-211 model at 60mA (rail voltage of 870) and the NVS-572 at 100mA. The NVS-003Gs use a push/pull OTL circuit and two mirrored power tubes may be removed to reduce output power to 35 watts per side. The two large transformers on deck feed the power supply and driver stage respectively. Again, there is no output transformer.

To circle the Navison wagon as it were, John Potis is presently listening to the trio of SE-MkI preamp and 20-watt NVS-211 monos. Owning Art Audio's 16wpc Carissa 845 SET and Canary Audio's 26-watt CA-330 push/pull 300B monos makes John the perfect critic to assess Navison Audio's standing in the mid-power SET sector, both from a sonic perspective and relative to build quality and value.

My forthcoming report on Navison's 75-watters will flesh out the emerging picture while I educate myself on what OTLs can bring to the party that even superior transformer-coupled SETs may not. General rumors about the genre indicate that one doesn't flock to OTLs expecting SET-type sonics and warmth. Rather, one pursues them for ultimate transparency, blazing speed, dimensionality of staging reputedly not achievable by other means and maximum directness.

Be that as it may -- I should soon find out -- $40 a pop for a Sovtek 6C33C-B originally designed for Russian MIG-25 Foxbat jet fighters is a far friendlier proposition than the alimony payments of supporting exotic direct-heated NOS triodes in upscale SETs. To retube all the power triodes here is $320 complete, about the price of a single good 45. Never mind that 45s, 2A3s, 300Bs and their ilk won't produce 75 watts of Class A juice unless one went seriously off the reservation with some monstrosity. As with any audio story then, there's promises and there's concerns. There's assets and liabilities and knowing how to maximize the one while minimizing the other.

With my Zu speakers, incoming WLM Diva Monitors, HornShoppe Horns, Gallo Reference 3.1s, Mark & Daniel Rubies with Omni Harmonizer and a surprise, I should have quite the contingent of test subjects to hit the OTL road with and see what is copasetic, what isn't and where general sonics will point...