Mark Levinson No.26 preamplifier & No.20 monoblock power amplifier Page 4

Rather than the ubiquitous 5-way binding post, Madrigal chose to use gold-plated Fischer sockets to connect the loudspeaker. These are rated at 50A continuous, 80A peak, and provide a gas-tight connection. (There is a satisfying "pop" when the plug is removed.) A color-coded pair of plugs is included with each No.20; your ML dealer should be willing to fit these to the cables of your choice, though Madrigal's own CPC cable can come prefitted. To aid changing cables, I used 5-way-to-Fischer adaptors for my tests, but I wouldn't recommend this for a permanent installation.

A matter of concern with an amplifier such as this, which has a considerable amount of stored energy, is what happens when things go wrong. Comprehensive protection circuitry monitors the following parameters: AC mains current; positive and negative rail currents and voltages; heatsink temperature; DC offset; power dissipation in the output stage, phase-angle-dependent; and whether the output is short-circuited. If a fault condition is detected, fuses, thermal breakers, and/or the relay-trip on/off switch hopefully shut things down. During the time I used the No.20s, which involved one inadvertent shorting of the outputs and one occasion when the No.26 was accidentally unplugged from the wall while an LP was playing, the amps seemed burstproof.

Madrigal stresses the importance of an appropriate warm-up period for all Mark Levinson products. At least one hour is recommended for the No.26; I left it on for 24 hours before doing any serious listening. Because of their profligate demands on the electrical supply, I didn't leave the No.20s on all the time, but in general turned them on in the morning to do serious listening in the evening.

The No.26 and No.20s were used both for my recreational listening and to drive various loudspeakers under review over an eight-week period. Source components used included a Mission PCM 7000 CD player (used both direct and driving a Sony DAS-703ES outboard decoder from its digital output), an up-to-date Linn Sondek/Ittok/Troika combination (with the latest composite armboard) sitting on a Sound Organisation table, an LP12/SME V/Koetsu Red player sitting on a RATA Torlyte stand, and a rather ancient but well-preserved Revox A77 to play my own master tapes. Loudspeakers predominantly used for serious auditioning were Celestion SL600s, placed on heavy, single-pillar Celestion stands spiked to the floor. True to my exposition of the relative importance of electronics and loudspeakers, you will note that the ratio of the respective prices approaches 10:1!

It could be fairly pointed out that use of just one loudspeaker may lead to idiosyncratic results; however, although I had originally intended to use B&W Matrix 801s in addition to the Celestions, circumstance dictated otherwise. As delivered, the 801s had defective tweeters (a problem due to an out-of-spec plastic molding and prevalent among the first US shipment). While I was waiting for the replacement tweeters, the speakers were stolen! That was that (footnote 4). Ultimately, I used the Celestions in combination with stereo Sumo Samson subwoofers (the main speakers being fed full-range) to get a handle on the No.26's low-bass performance. In addition, at the end of the review period, I took delivery of an intriguing pair of English loudspeakers which positively sang when driven by No.20s. The Acoustic Energy AE1 combines a version of the ELAC aluminum-dome tweeter used in the Monitor Audio R952MD with a 5" aluminum-cone woofer, and possess exceptional lower-midrange clarity.

Due to the ML gear's use of Camac and XLR connectors, and a limited supply of adaptors, interconnect choice was limited. I used Madrigal's new HPC (Helical Planar Copper) interconnect (footnote 5) both for connecting source components and for connecting the No.26 to the No.20s. I tried the latter both in unbalanced and balanced fashions. I also carried out some preliminary balanced auditioning with solid-core, twisted-pair RS232 cable (the only symmetrical cables I had that were long enough). Differences were subtle, but ultimately I felt that the balanced connection using HPC had the most authority across the band. For connection of the speakers, I first used Monster M1 cables, then switched to Madrigal's CPC (Co-Planar Copper) solid-ribbon, Teflon-insulated speaker cable. Less fat in the midbass than the Monster, and a little more incisive in the treble, this gave a more even tonal match with the Madrigal interconnect.

How did the ML components sound? First, a minor digression: You may have noticed that a minor debate has been raging recently in Stereophile concerning whether the "accuracy" and "musicality" (for want of more precise language) of a component are mutually exclusive. In his "Final Word" column last month, LA commented that "if accuracy isn't musical, what does that say about what we're doing in and with high fidelity?"

In my opinion, "accuracy" or "neutrality" is something that designers strive for but nearly always fail to achieve, particularly regarding preamplifiers, which fall short in one of two major ways. The first is when the component is not particularly neutral, but still delivers a fair degree of musical enjoyment, despite the detail being smeared over to one degree or another. Based on my experience with specific models, into this camp fall such UK models as the original Naim NAC32, the Exposure VII/VI, and Ben Duncan's DIY AMP-01 design (published in HFN/RR in 1984). In the US, the Conrad-Johnson PV5, Counterpoint SA-7, Audio Research SP8, and (on a higher plane of achievement) the ARC SP10 and Conrad-Johnson Premier Three, all fit into this class.

Footnote 4: The serial numbers of these speakers were 555 and 556.—John Atkinson

Footnote 5: This is a low-capacitance design with wound "solid-core" flat conductors for both signal and ground. The shield doesn't carry any current and is connected at one end only. (Arrows printed on the insulation point away from this end.)—John Atkinson