Cary Audio Design CAD-805 monoblock power amplifier Sam Tellig 3/98 part 2
Don't you just hate it when that happens?
If you don't hate it, you may love single-ended triodes.
One of the great joys of discovering single-ended triodes is that you may start enjoying certain kinds of music that you find largely unlistenable with non-SET triode amps. Take chamber music, for instance.
"Yes, take it," you say. But wait.
If you avoid string-quartet recordings because they sound sharp and edgy, even wiry, try a single-ended triode amp and think again. You may find you prefer listening to chamber music than to large-scale symphonic works. After all, chamber music is scaled for reproduction in the home; the Vienna Philharmonic is not. It's no coincidence that Dennis Had's father was a violinist and played professionally in a string quartet.
You may find that vocals are much more pleasing. That you can even enjoy opera.
As for the SET amp's measurements, what are you worried about? Are you going to measure your amp...or listen to it?
"I could have made the CAD-805C measure better," a cheerful Dennis Had told me. "But when I tried that, the magic went out the window."
Maybe one big reason single-ended triode amps tend to sound as good as they do is circuit simplicity—no, not Circuit City, circuit simplicity. The simpler the circuit, the fewer parts you need. The fewer parts, the purer the sound.
"I have the ultimate single-ended," trumpeted Valery, my friend from Minsk. "Single-ended no-ode. Beats even triode."
No-ode is Val's windup gramophone and prized collection of 78rpm recordings.
"Single-ended, just like in nature," said Val, as he wound up the gramophone between the Russian salad and the grilled sturgeon. (Val wheels the gramophone over to the dinner table. It makes for busy meals.)
Nipper did indeed hear His Master's Voice...and it was single-ended. You think I'm kidding, but you should listen to a windup gramophone someday. There is an immediacy to the sound, a living presence—oops, that's been used—that electronics of any kind appear to compromise.
What is this single-ended business, anyway?
Basically, single-ended amps—and these can be solid-state, as with the Pass Aleph amplifiers—amplify the positive and negative phases of each waveform continually, with no phase splitting between the positive and negative phases as there is with a push-pull amp. This is what "push-pull" means. Push amplifies the upside phase, then pull takes over on the downside. It's the kind of thing that appeals to engineers because it looks so clever.
Some very talented amplifier designers have told me they can devise phase splitters that have no deleterious effect on the sound.
Perhaps. I've heard some very good push-pull tube amps, some of them using the 300B triode as an output tube. Push-pull makes for more amplifier efficiency—you use output tubes in pairs for more power. And push-pull avoids the problem of output-transformer core saturation.
Still, common sense tells you that something about push-pull appears to violate nature. It can't possibly help the sound. In my opinion, it doesn't. To my ears, push-pull does something to the harmonic presentation—to sonic purity, if you will—that you don't notice until you've become accustomed to single-ended.
"Why does hi-fi sound so different from live music?" a member of the "Ask the Editors" audience at HI-FI '97 in San Francisco wanted to know.
"You wouldn't ask that question if you owned a single-ended amp," I quipped.
Other Stereophile personnel on hand were having none of this. It's safe to say that no hi-fi can sound like live music.
Me? I'm dangerous. I'm saying precisely that single-ended triode gets you so close to "live" you could swear that Louis Armstrong, Billie Holiday—hell, Sviatoslav Richter—have all been raised from the dead.
I wasn't around when talking pictures became all the rage, but my parents were. Perhaps one of the reasons people were bowled over by talking pictures—and early radio—is that the sound was all single-ended. The engineers hadn't yet had the opportunity to ruin reproduced sound with push-pull.
All right, maybe I'm going too far.
But I think single-ended triodes may really take off when they're used in home theater. For this reason, I've been after Dennis Had to produce a relatively inexpensive, 300B-based mono amp, to be sold in threes—left, right, and center—for home theater. You have not heard great home theater until you have heard it in single-ended triode.
I am not joking.
Hear no evil
There's another evil that single-ended triodes generally avoid: negative feedback. There seems to be a growing consensus that negative feedback is bad for sound. Again, common sense should tell you that it is. Negative feedback adds distortion. It has to. You're putting signal back into the signal—creating an endless loop of low-level grunge.
I'm not the only one who's become distrustful of negative feedback. Here's Martin Colloms: "There is a growing suspicion that some...aspects—a loss in natural timbre; a duller, less expressive performance; increased aural fatigue; and missing life and energy in reproduced sound—may be consequences of the application of negative feedback." (Martin Colloms, "A Future Without Feedback?," Stereophile, January 1998, Vol.21 No.1, p.89.)
Why do designers add negative feedback?
"The only reason, in my opinion, for feedback at all is to lower the output impedance," declares Dennis Had. "Now you're forced to use negative feedback if you use output tubes in a pentode or tetrode configuration because that extra element that's put in the tube, compared to a triode, is generating odd-order harmonics.
"If you use the tube in the triode mode, you won't have to clean up the problem. That's why I am such a big proponent of running the KT88s and the EL34s in triode mode in our push-pull designs. The harmonic garbage is gone in triode mode. Then I add just a tad of negative feedback to lower the output impedance.
"With our push-pull amps, you can stick them on any speaker and you don't care. With the single-ended triode amps, using no feedback, you have to be more cognizant of the kind of loudspeaker you're running it with."
With no feedback dialed in, the output impedance is slightly less than 2 ohms, according to Dennis. That's certainly high enough to bother some measurement wonks (footnote 2). And true, you will likely get frequency-response anomalies if your speakers' impedance is all over the place.
One rule of thumb with high-output-impedance single-ended designs is to pick a speaker whose impedance curve is relatively flat—in other words, a speaker whose impedance stays close to the nominal 4 or 8 ohms across the frequency spectrum. If the impedance varies by more than 20% or so, up or down, you may have problems. Thus, with a nominal 4 ohm speaker, you wouldn't want the impedance to dip below 3 ohms or rise above 5 ohms. With a nominal 8 ohm speaker, you wouldn't want the impedance to stray beyond 6 to 10 ohms.
With the Cary CAD-805C, you can dial in up to 10dB of negative feedback and thereby lower the output impedance. But listen to what it does to the sound! The harmonic presentation suffers. The extraordinary holographic soundstaging is compromised. As Martin Colloms noted, the more feedback you dial out, the more of the amp's magic you can hear. You can bring the amp into focus—and "on song," as the British like to say—simply by twiddling the feedback knob. In one direction. Down.
Once again, amplifier designers swear up and down that they can "correctly" apply negative feedback so that sound quality doesn't suffer—for instance, by applying a little local feedback at various stages of the amplifier circuit rather than a gross amount of global feedback.
Footnote 2: I can't repeat this too often: Music lovers suffer at the hands of audiophiles, especially those who take measurements. Remember, they've got to measure something. That's their job.—Sam Tellig