Conrad-Johnson ART Preamplifier Interview

To Know a Thing by its Parts is Science;
To Feel It as a Whole is Art

Wes Phillips talks with Lew Johnson about ART and engineering

Lew Johnson: The ART was not originally conceived as a product---it was a research tool we built for ourselves. We realized that much of what we had learned in building power amplifiers had implications for preamplifier design as well---we came to see that loop feedback is not innocuous, and that local feedback has its own drawbacks as well. We wanted to design a preamplifier that came as close to being a zero-feedback amplifier as possible.

At the same time, we wanted to go to even further extremes in fitting out the power supplies with polystyrene capacitors instead of polypropylene. We conceived the idea of eliminating cathode followers by constructing a fairly extensive composite triode, where you take multiple triode sections and connect them in parallel---which gives you one high-transconductance triode.

Wes Phillips: Doesn't using the tubes in parallel minimize noise? Johnson: You get a remarkably good signal/noise ratio because, in theory, you improve SNR by 3dB every time you double the number of triodes. However, while [paralleling the tubes] minimizes thermal noise, it doesn't do anything about other types of noise---microphonics, for example. If you get a microphonic tube, the fact that there are 10 parallel triodes doesn't help.

Phillips: What do you estimate tube life to be for the ART?

Johnson: Our experience so far suggests two to three years averaging 10 to 15 hours per week---about typical for our gear. Because a line stage doesn't deal with huge amounts of gain, a tube has to get really noisy before it's noticeable.

One thing you want from a good preamp is low output impedance, so it doesn't interact with the interconnect cable between the preamp and the amp. This is the fundamental flaw in so-called passive line stages: the volume control is in series with the interconnect cable and the volume control is a relatively high-impedance source---at least, it can be, depending upon its setting. Designers usually incorporate a buffer stage into the preamp. In the case of a tube preamp, that's a cathode follower---which is a device that has a high input impedance and a low output impedance and a gain factor of, essentially, one (actually it has a very slight amount of loss).

Our idea was, "Let's get rid of the cathode follower by constructing a sufficiently butch triode that will lower the output impedance down to where we want it." In the case of the ART, [the impedance] is a little below 500 ohms; in a typical tube preamp, output could fall anywhere between a couple of hundred ohms and 500 ohms---the ART's in the ballpark. You may be able to find cables that would present a problem, but most shouldn't.

We built that preamp and it remained in our factory reference system for seven or eight years. The combination of all those triodes and other associated bits and pieces, along with the extensive polystyrene power supplies, made for what would have to be a very expensive preamplifier---and, frankly, we just didn't see where there would be a market for it. The word "reference" in its name is not marketing hype; we built it to use as a designer's reference when we worked on other products.

The audio circuit itself couldn't be simpler: It's a single triode. I don't want to go into details about biasing the tubes, but there is absolutely zero local feedback in this preamplifier---and no global feedback either. Most self-bias circuits automatically have local feedback---or have a large capacitor in the circuit, typically electrolytic, which has its own evils, as you can imagine. We use a different approach that eliminates both the feedback and the capacitor.

The control circuitry is essentially a computer that we use to set switches and has no relation to the audio circuit. The beauty of the computer is that it allows us to build a completely dual-mono preamplifier that has a single set of controls. A control link between the two chassis allows the user to have a single control interface with the unit, but it can command switches resident in each channel's circuitry.

X