Hindsight. When penning my Standouts of 2018, I'd nominated the €30'000 Allnic Audio L-10000 as my probably most expensive assignment of the year. It recalibrated personal cynicism about what's possible in the preamplifier realm. I wrote, "it opens up a new way for creating a top system which runs 180° out of phase with thinking that quality 64-bit dithered digital volume controls have eliminated the entire rationale of the preamp breed. The Allnic direct-couples balanced-drive 300B to your amplifier without capacitors, transformers or audible noise. By not driving reactive low-impedance speakers but the stable high input impedance of an amplifier, the direct-heated triodes breathe easy. By not working through a bandwidth-limiting phase-shifting transformer or time-smearing capacitor, they express the inherent triode aroma in full. Balanced drive and unique triode/pentode inputs offset typical 2nd-order THD effects.
"The upshot is that in conjunction with superior transistor amps like LinnenberG's Liszt driving big traditional multi-way speakers, the L-10000 can inject uncontaminated perfectly stable direct-heated triode aroma without sacrificing low-impedance drive or bass control on rhythmically dense infrasonically charged modern music. One gets all of the tube gain without any of the pain. One goes places where small 12AU7/6922-type preamp tubes can't; where even DHT amplifiers won't go unless matched to a rare truly copasetic loudspeaker but still fed a select rather than omnivorous music diet."
Truly, the mix of DHT in a line stage with transistors on the speakers made for a terrifically convincing best-of-both-worlds scenario. With the Allnic, it represented an idealized hybrid workaround for those of very deep pockets who wanted DHT sound but wouldn't tolerate its limitations when strapped directly to loudspeakers. Typical limitations are bandwidth compromises in the extremes and as a result, a tell-tale midrange emphasis; compromised bass control; congestion on dense dynamically challenging music; a shift away from taut transients into a more reverb-heavy decay-rich milieu; and lowered resolution from higher operational self noise.
To generate its true-balanced i/o around a single-ended circuit, the L2 uses active line drivers where the L-10000's gain circuit itself is balanced.
Today's Vinnie Rossi L2 fits the same 'build a perfect hybrid' bill but for half the coin. It also omits Allnic's far more complex three-stage circuit of 6AN8 inputs, 12AU7 drivers and push/pull 300B which total eight tubes but clock a vintage -80dB S/NR. With the L2, owners get to have 21st-century performance specs and their DHT of choice without residual microphonics or sonic course adjustments from valves other than that single power triode per channel. I propose that to hear what any of these direct-heated power triodes really sound like is possible only with such a spud circuit. It eliminates the fingerprints of the usually contributing parts. In Manley's DHT preamp for example, 6SL7 precede their 300B. Four more ancillary tubes show up as 5U4G rectifiers and OD3 regulators to increase maintenance. The Coincident 101D preamp uses output transformers, the Supratek 6SN7 drivers again. If things were as simple as they are ultra minimalist in the L2, we'd see more such designs.
Regulations. We're already keyed in why DHT preamplifiers remain so rare. It is tough to get the noise of their inherently microphonic direct-heated output triodes on par with modern expectations – should those be informed by the very best digital sources which have begun to cross the -130dB line. Obviously some noise behaviour points right back at particular tube types and makes which differ in mechanical makeup and microphony.
Reading the L2's web page, we come across repeat mentions of custom Belleson super regulators. Like equivalent Teddy Pardo and Walt Jung parts, Brian Lowe's now patented voltage regulators claim better ripple rejection and noise performance than cheaper commercial alternates.
For our purposes it suffices to remember that the decibel scale is a logarithmic function, then look at Belleson's published comparison charts for a simple appreciation that the specs of different voltage regulators span quite the gamut.
Which part for a given application is best always depends on the designer and his/her cost allowance. As these graphs illustrate, ripple-rejection effectiveness varies across frequency. A given part's ultimate usefulness depends also on whatever bandwidth a circuit considers most critical for noise suppression.
With his statement ambitions for the L2 meant to eclipse LIO's pre-existing DHT option; and his most critical noise spectrum being the 50/60Hz power line with its lower harmonics… Vinnie obviously felt that his custom Belleson super regulators were the best possible option to secure the very low noise he was after.
The higher the number on the vertical dBV scale, the higher the noise rejection.
Here we see the operational noise floor of sundry voltage regulators, with the SPJ78 once again the Belleson part of the comparison.
Two Belleson SPZ 3-ampere super regs with 110dB ripple rejection at the power line frequency also show up in the optional L2 phono module. Ditto as main regulator for the L2 DAC. That supports up to 32/768kHz PCM and DSD 512 with displayed sample rates. Converters are one mono AKM AK44497EQ per channel. An FPGA with femto clock handles jitter suppression. The 'filter' button on the remote can set non-oversampling mode with digital filter bypass or minimal-phase filter.
The remote's 'phase' button inverts absolute polarity. More reed relays handle track changes between different sample rates to suppress switching transients. A total of three digital sources may wire up via USB, 75Ω BNC (RCA adapter included) and Toslink, then be toggled by remote. The DAC outputs 2Vrms with a S/NR better than 115dB. The DC-coupled 4-layer L2 phono board gets its own threesome with one MM and two MC inputs. The remote can tweak cartridge loading from 10Ω to 1kΩ on the fly which gets displayed. Internal jumpers adjust phono gain which is generated by four cascaded stages of 20dB/ea for "higher bandwidth and lower phase shift". At 3Vrms, the noise spec for vinyl exceeds -90dB.
Because the preamp's output stage takes so many different tube types, its precise voltage gain will vary even between brands but is approximately 9dB to not 'gain poison' your system. Should a tube tire or just for a sonic change of pace, switching to an ultra-wide bandwidth class A Jfet buffer by internal toggle removes the triodes from the signal path and kicks in 'passive' mode for zero voltage gain. This doubles as nifty neutral reference to determine what exactly the posh triodes bring to the party. In either case, the L2 volume control is a 64-step resistor ladder switched by silent Pickering relays to avoid clicks between level changes. Internal again are adjustments for the front panel display and LED brightness.