After putting 100 hours on Luca's glass, 'twas time to compare it and his circuit to Vinnie Rossi's grounded-grid effort with its direct-coupled—no capacitors, no transformers—L2 Signature. I fitted mine with my favorite Elrog ER50. Because thrice the power had sounded even better than Bakoon's AMP-13R, I'd by then already replaced that with Crayon Audio's CFA-1.2. Having exchanged emails with its retired designer Roland Krammer in Budapest a while back, I'd learnt that his current-feedback circuit exhibits exceedingly low THD. Its faint remnants of the 2nd and 3rd harmonic are equal. Neither dominates. That now seemed like a good tabula rasa for my tube preamp experiments. Unlike our Nagra Classic's 12AU7 had with its negative feedback against the big 7-volt ER50 bottles, Luca's small triodes didn't really sound drier. But they did project the soundstage more forward and rendered its objects more robust, firm and earthy where the DHT played it more set back, airy, spacious and temporally elastic.

An ideal demonstrator track for this difference was "Sacred Mind" from the Lars Danielsson album Cloudland on the premium ACT label. It's an inherently dreamy tone poem much like an impressionist Monet painting. The big Teutonic triodes revealed lengthier trailing edges behind the piano, around the moody trumpet, chasing the opening's strange rippling on each bar's third beat or enhancing Kinan Azmeh's questing clarinet behind the piano's waterfall figures during the bridge.

Luca's small twin triodes minimized the same reverb action. That gave tonal cores more significance than the Vinnie Rossi which was even more expressive with their auras. This reminded me a bit of how the morning sun quickly burns off lingering night-fog ghosts to banish their mysterious half light. Soon everything deals in robust light-of-day facts. This difference wasn't really about tonal balance but more about attitude or perspective. The Vinnie Rossi presented things a bit more feminine so more supple and vulnerable. The Riviera acted more masculine, concrete and self assured. Of course applying living traits to inanimate objects is a slippery slope. Where it conveys a recognizable feel during playback, one just trusts that the audience appreciates the intent and sees its inherent limitation in the very next breath. In more audiuophile terms, the big triodes showed extra energy in the air region but no parallel brightness like a good super tweeter that just enhances spaciousness. If there was any extra energy in Luca's preamp, it presented in the lower midrange. I phrased this as plausibility to suggest why perhaps it sounded meatier and more substantial. If Vinnie Rossi's was the spatializer, Luca Chiomenti's once more meant materializer. This contrast was a bit like one might expect from a high-gain versus low/no-gain circuit. It affected from which perspective I looked at the music. With the American it sat a bit farther off. It was surrounded/embedded in more overt space thus more reverb gossamer. Over there. With the Italian some of that space dried out. Rather than extend to anything resembling dryness alas, its color temperatures and image density increased my perception of in-room embodiment. Right here. Minus the former's very useful display and extra €5'000 on the bill then, I found the APL10 to perform on the same plateau. It held membership in the same club but managed entry with €70/pr glass, not €1'275/pr… a happy Marxist comeuppance within undeniable bourgeoisie. After all, we are racing on a very elite track here no matter how affordable this vacuum fuel might be. Still, such twists do deserve applause. One could roll from 15-20 different pairs of valves in this deck for the price of what just one pair gets in ours. So there.

Luca: "Some details on our Riviera preamps. There are some common choices which you already identified. One of them is the SRPP or Totem Pole circuit. This is not to be confused with circuits like the SEPP (single-ended push/pull) or the mu-follower. They all belong to the same family but differ in their behavior and sound. SRPP was developed for radar output stages and the first known audio uses date back to Japan of the 1960s which Jean Hiraga introduced to the West in the '70s. There are many examples of it—I started to use it in the eighties—but I'm not sure that everyone who used it really understood its behavior. The SRPP is a strange kind of push/pull. It is a true push/pull only if there's a particular relationship between load and polarization. Any good technician will do his best to realize that condition to obtain all the benefits of a real push/pull circuit which is to reduce even-order harmonic distortion. But are we good technicians? Perhaps but surely we are more interested in the musical than measured result. So what happens if the perfect push/pull condition in a totem pole circuit is not realized? Even-order harmonics don't cancel. If one knows how to model this behavior, it becomes that of a single-ended circuit just with less distortion. It all depends on the tuning even if technically it remains a push/pull circuit.

"Also, a standard preamplifier sets volume and balance with two potentiometers in series. Adopted by 99% of manufacturers, this has some issues. When not at zero or low levels, a pot has variable usually high output impedance and always a fixed input impedance. To have good drive over the following stage or pot, the first pot should be a low-impedance type which is a poor choice for the input. The second pot has to be of high impedance to not load the first but this is a bad choice for the output of the control section. A buffer stage with unity gain could insert between the pots but means another stage in the signal path with all the usual problems. So the going solution is a compromise between the chosen values of the cascaded pots. We beg to differ by moving one pot to the output. Volume remains on the input, balance sits at the output. As you can imagine, this solution has its own challenges. In the APL01 we use two separate balance controls, in the APL10 a single control with precision voltage divider switched by reed relays. I find this solution extremely satisfying from both the technical and musical points of view.

"Another common point for our preamps is that our single-stage circuit inverts phase. This has been subject to long discussions over the years. I don't think it's a real issue because we never know the recorded phase. About half the recordings are in phase, the other half is out of phase. Also, many power grids invert the polarity of connected power supplies. I always suggest to test your power polarity by reversing the power plug [only works with Schuko plugs, not US or UK variants – Ed.]. Now we get to the differences between our preamp models. The APL10 uses ECC82, the APL01 6350 or 6463. Then their power supplies differ not just in size but structure. The APL10 uses a resistive double-PI filter, the APL01 an inductive filter. The APL10 has a very well-implemented series regulator, the APL01 an exclusive shunt regulator. The volume pot in the APL01-SE is a motorized TKD, in the APL10 it's a motorized Alps. Finally the APL01-SE uses output transformers for both its balanced and unbalanced outputs.

"The signal path of the APL10 is extremely simple. Between the i/o ports, it's the input switching relay ⇒ the quad Alps pot with two sections in parallel to improve performance and tracking ⇒ the tube grid of the first tube ⇒ the SRPP stage ⇒ the output capacitor in a good-sounding mix of parts ⇒ the balance pot. It may look like very little but I believe that we obtain the best results with a few well-chosen parts that were applied with proper knowledge and intelligence. The technical specs are what you'd expect from this type circuit. Bandwidth is ~6Hz-500kHz -3dB, quite impressive for a zero feedback circuit. I'm not fanatical about ultra-wide bandwidth but with an SRPP that 'comes for free' and I prefer not to limit it. Input impedance is a standard 50kΩ, output impedance ~1.2kΩ which won't have issues with any power amp of a 10kΩ or higher input impedance. Voltage gain is about 6dB so adequate for today's high-gain D/A converters. Circuit noise with no feedback is influenced by actual tube performance where a noisy tube can modify the result. The absolutely worst case I ever measured was about 75µVrms unweighted so an SN/R of about -83dB unweighted referenced against a standard 1Vrms output. The weighted value is well over -90dB."