Mark Levinson No.26 preamplifier & No.20 monoblock power amplifier Page 3
Mark Levinson No.20
I first heard the Mark Levinson HQD loudspeaker system some ten years ago, at the Festival du Son in Paris, in the days when it was a real hi-fi show. Perhaps even more so than by the speakers, however, impressive as they were with their stacked Quads, Decca ribbon tweeters and Hartley 24" woofers, all in beautiful beech woodwork, I was bowled over the amplification used: six Mark Levinson ML-2 monoblock power amplifiers. Capable of just 25W output each, these class-A monsters were the first true high-end amplifiers I had ever heard. The memory of effortless low-frequency reproduction, coupled with a magically liquid midrange, lingered on.
Then, some two years ago, the first Mark Levinson product to emerge from Madrigal Audio Laboratories was a modern recasting of the ML-2. Sharing its predecessor's physical size and appearance, the No.20 offers not 25 but 100W into an 8-ohm load, this doubling as the load halves down to 2 ohms, indicating that the peak output voltage of 40V (28.3V RMS) doesn't droop into the lower-impedance loads. This is true voltage-source operation. Like the '2, the No.20 is a class-A amplifier, offering up to its 100W rating into an 8-ohm load without either npn or pnp output transistors turning off. (Into lower impedances, the same peak voltage requirement will turn off upper or lower transistors, giving class-AB operation on such peaks.) Although true class-A operation results in total freedom from crossover distortion or thermal modulation of the output transistors' transfer characteristics, the penalty is higher running costs: each No.20 sucks 500W from the wall when sitting, quietly minding its own business. (This is equivalent to almost 5A; prospective No.20 owners should ensure that they will not have any other current-hungry appliances on the same mains circuit. The circuit-breaker might not pop, but it is probable that the sound quality will be compromised.)
The power supply is larger and more sophisticated than you would expect for an amp of this rating: after RF filtration and spike protection with large MOVs (Metal Oxide Varistors), the AC line feeds a separate 600VA toroidal transformer/35A rectifier bridge/24,000µF filter combination for each unregulated voltage rail, positive and negative. Three traditional ML-style heatsink clusters line each side of the No.20 chassis; the pair nearest the front carry the series-pass bipolar power transistors used to regulate these rails; neither driver stage nor output stage is therefore perturbed by voltage fluctuations which would otherwise result from heavy current demands. As the impedance of the secondary of the line transformers is in series with the output transistors, the higher the current drawn from them to recharge the filter caps, the lower the voltage they can supply. Regulation increases the steadiness of the voltage rails by a factor of up to 60dB or so, depending on the topology and transistors used, but is rare among amplifier output stages. Essentially, it demands a second power amplifier, delivering current at a constant DC voltage, upstream from the audio amplifier, which considerably increases the basic cost. Following the regulator, each of the No.20's regulated rails is then further filtered by a 9000µF capacitor. All the electrolytic caps are bypassed with plastic-film types.
The No.20 has two sets of inputs: inverting and non-inverting via Camacs, and a balanced input via a Neutrik XLR socket, offering three modes of operation. If the amp is required to be used unbalanced, a shorting Camac plug (supplied) is inserted into the socket not used to connect the other "half" of the balanced input to ground. Input impedance is moderately high for a solid-state design, at 50k in parallel with 1nF.
All the active circuitry is contained on plug-in circuit boards, using gas-tight, mil-spec, Varicon edge connectors to facilitate easy service. The input and voltage-gain circuitry are carried on two boards near the unit's rear and consist of a differential amp, both halves of which are driven in balanced mode. This in turn drives a second differential amp, which includes a current inverter to convert from differential to single-ended signal handling while maintaining push-pull operation. This stage also includes a type of soft-clipping circuitry: the amplifier's open-loop gain is reduced if it appears that the signal is nearing a level liable to drive either this stage or those downstream into clipping. The driver stage is compound, a complementary emitter-follower driving in turn three more complementary emitter-followers in order to get sufficient current gain to fully drive the output stage. This consists of eight complementary push-pull emitter-followers in parallel, two pairs of bipolar power transistors occupying each one of the remaining four heatsink clusters.
Like the Krell and Naim amplifiers (and the Hafler XL-280), the No.20 lacks the conventional series output inductor used to define a low-pass pole with the cable capacitance and thus ensure unconditional stability with an amplifier that might otherwise be a little doubtful in this respect. This output inductor both raises the output impedance at high frequencies and is thought by many to be a source of audible ills. All voltage-gain and driver stages are said to run in class-A, and the amp is DC-coupled throughout.