Mark Levinson No.26 preamplifier & No.20 monoblock power amplifier JA September 1989

John Atkinson wrote about the Mark Levinson No.20.5 in September 1989 (Vol.12 No.9):

Some things are down to timing. When Madrigal's Mike Wesley visited Santa Fe just before Christmas 1988 with this revised version of the Mark Levinson No.20 power amplifier that I had reviewed back in Vol.11 No.5, I was in the middle of listening sessions for a number of loudspeaker reviews. I did not wish to insert a radically different power amplifier into my system halfway through those tests. Although I did plan to write a review of the amplifiers, at that time it seemed appropriate, therefore, that the pair of No.20.5s should spend some time in Larry Archibald's system, particularly as he had been using the No.20 with the No.26 preamplifier for the review of Thiel loudspeakers that appeared in the January 1989 issue of the magazine.

Well, time passed; it was the end of February. I had finished my reviews of inexpensive loudspeakers and introducing a significant variable in the form of new power amplifiers would not be a problem. "Uh-huh," was LA's reaction. "I'm going to need the Levinsons for the review of the Mirage M-1 loudspeaker in the June issue." That was OK, I had a number of projects that I was about to embark upon in which it would be a good idea to keep the amplification chain constant. I would listen to the Levinsons at a later date.

Time passed; it was the beginning of June. I asked Larry if I could come round and get the 20.5s. "Uh-huh. I'm going to use them to drive the Quad US Monitors." Now this is very untypical Archibald behavior. There must have been something about the Levinsons that was proving addictive.

There was. It was their sound—or rather, their lack of any particular sound. Whatever loudspeaker Larry tried with them, though still imposing its own signature on the sound, the 20.5s seemed capable of producing a sound that always remained true to the needs of the music. In fact, during a listening session at Larry's that featured the humongous—and horrifyingly poor—Altec BIAS 500s, I joshed that even the poorest pair of speakers we could lay our hands on would still sound at least acceptable when driven by the Levinsons. Accordingly, we set up a pair of single-unit squawk boxes that had accompanied a "vertical boombox" that Larry had bought for his daughter Rachel. (Yes, he does love her, but she had wanted something identical to the systems that her friends owned—"Not hi-fi separates, please, Dad" was the instruction. Larry almost complied: he felt that replacing the truly appalling supplied speakers with the surprisingly musical Boston Acoustics A40s would still conform to the letter of Rachel's requirements.) We set the squawkers up and connected them to the Levinsons.

Well, they were still dreadful: rather than a continuous frequency response, the sound consisted of a series of isolated resonant peaks with nothing in between. Yet...they weren't as unlistenable as we'd expected. You could even hear flashes of music.

Despite Larry's protestations I removed the Levinsons from his listening room and set them up with the Martin-Logan Sequel IIs. Before going into the details of my listening impressions, it would probably be a good idea to briefly discuss how the 20.5 differs from the original No.20.

Design & development
The No.20, the first new Mark Levinson product from the company since being taken over by Madrigal, was introduced in early 1986 and, considering its $10,000+/pair price tag, has sold in what I understand to be relatively high numbers. Its rather reticent balance was more to my taste than the more upfront presentation of the later No.23. Design-wise, however, the 20 could be best regarded as an evolutionary development of previous Mark Levinson products. For example, the manner in which transistors on its AP-3 input and driver printed circuit board were thermally coupled would have been familiar to any owner of an ML-7A preamplifier. As a result of the development of the Mark Levinson No.26 preamplifier and No.23 power amplifier, however, the Madrigal design team has worked on better methods of ensuring good thermal tracking (as well as increasing the circuit's basic linearity by, among other things, making use of cascoded bipolar transistors). Back-engineering similar improvements to the No.20 design has resulted in a new input and driver card, the AP-4, which is a plug-in replacement for the AP-3 and solely distinguishes the No.20.5 from a late-production No.20. (This card also carries an improved controlled overload circuit that shapes the waveform at the clipping point so that its content of high harmonics is minimized.)

Owners of No.20s can, therefore, have their amplifiers upgraded to No.20.5 status by having their Mark Levinson dealer change the card approximately for the price difference between the two amplifiers—$1000/pair—though amplifiers with low serial numbers will require additional parts and labor to modifify the six output-stage modules. (All the cards are connected to the mother board via gas-tight, Mil-spec, Varicon edge connectors.)

A full description of the No.20's circuitry and construction was included in my review in Vol.11 No.5; as the No.20.5 only differs in detail, I don't see that I need recapitulate that description other than to touch on the main points. The power supply consists of separate 600VA toroidal transformer/35A rectifier bridge/24,000;mF filter combinations for each unregulated voltage rail, positive and negative, with—unusual in a power amplifier—series-pass bipolar power transistors then used to regulate the rails to the bipolar output devices, which are run in class-A. As should be mandatory with an amplifier at this price level and with this level of stored energy, full protection is provided, and the No.20.5 is said to be burst-proof.