Threshold SA-1 monoblock power amplifier

Eleven years ago, Threshold Corporation entered the high-end audio market with the first amplifier ever to use sliding bias (footnote 1) in its output stages. Some 10 years later, Threshold spawned another innovation: their so-called Stasis circuitry, which yielded the S-series amplifiers. The SA-1 and its lower-powered sister SA-2 are the latest from Threshold, and are the first Threshold amps to abandon sliding bias for straight class-A operation. Both use the Stasis circuit.

In most amplifiers, the voltage amplification stages are used to drive the output transistors, which in turn drive the loudspeaker. In a Stasis design, the voltage amplifiers connect directly to the loudspeakers.

Obviously, this cannot work. If you recall Ohm's law, or even if you never knew it, you may take my word that current equals voltage divided by resistance, and power equals current squared times resistance. That means that any resistance across the output of a voltage amplifier will require that amplifier to deliver power as well as voltage—something it is better off not being asked to do. The ideal load for a voltage amp is an open circuit—infinity ohms. The Stasis circuit allows the loudspeaker to behave like just such an open-circuit load.

Normally, any attempt on the part of the voltage amp to develop an output voltage would be stymied by the current drawn off immediately by the loudspeaker load. Virtually no voltage would develop. But backing up (as it were) the SA-1's voltage amp are two banks of current-amplifying devices—output transistors—and a proprietary current-sensing circuit between the voltage amps and the loudspeaker. Whenever the sensor detects current being drawn from the voltage amplifier, it gates open the output transistors, which dump in as much current from the amp's power supply as is necessary to bring the signal voltage across the speaker up to the same level that the VA is trying to deliver.

For example, let's say the VA is trying to deliver 5 volts. Before the speaker's current drain has time to deplete this, the output transistors will dump enough current through the speaker to bring the voltage across it up to 5 volts. The voltage amp, seeing 5 volts at its output, will then deliver no current at all, just as though it were working into an open circuit.

The current gating must only be able to respond faster than the delivered voltage changes, in order to keep "on top" of the ever-changing current demands. As long as it does, the voltage amp will "see" what is essentially an open-circuit load. Thus, since we define work in terms of power consumed, it can be fairly stated that the Stasis voltage amps do no work at all, which greatly simplifies front-end power supply design. Devilishly clever, wot?

The SA-1's input stage consists of four matched junction-type FETs, with the first pair operating as a differential pair and the second pair in a cascode arrangement. Because JFETS have an inherently very high input impedance (unlike bipolars, which draw current from the input signal), it is unnecessary to play games with feedback or "bootstrapping" (footnote 2) to minimize the load across the preamp outputs. There is consequently less chance for adverse interactions between the preamp and the power amp. In addition, the front-end elements are so arranged as to behave complementarily in response to power-supply voltage variations, making the front end virtually immune to signal-induced supply fluctuations stemming from the high-current output stage. Not that such fluctuations should be that much of a problem here: The SA-1 has a humongous toroidal power transformer and an incredible 120,000µF of power-supply storage, suitably bypassed with film resistors.

The output stages look like overkill, too. Threshold claims that each output device is rated by its manufacturer at 20 amps at 200 volts, which gives each one a power capability of 20 watts for every watt of rated power! That's called conservative design; it allows the designer to dispense with "protection" circuits, which nearly always befoul the sound when they're not, occasionally, saving the amplifier.

The SA-1 has no overall feedback, no output fuses, and no inductor in series with the output (to protect it against very low impedances at very high frequencies). These things allow it to present an extremely low source impedance (high damping factor) to the speaker, which no doubt contributes—along with its very high current capability—to its remarkable bass performance, about which more later.

Do I hear a question from my audience? Ah yes, a good point. Isn't the SA-1's price/power ratio rather poor? Well, at $6000 for two 160W channels, that calculates out to $18.75 per watt, which is lousy. (The 140Wpc Electron Kinetics Eagle A, at $960, costs $6.85 per watt; the Adcom GFA-555 costs $3 a watt. Only the Mark Levinson ML-2 offers substantially lower $/Watt value than the SA-1, at an unbelievable $176 a watt!)

Lately, however, I find myself questioning conventional wisdom about requisite amplifier power. Most audio perfectionists feel that 200 watts per channel is the minimum needed for clean crescendos. But 160W will make only a 1dB difference in the maximum level the system can deliver cleanly, and a 1dB change in across-the-board output level is barely perceptible by most people, if it is perceptible at all. The subjective difference between 200W and 160W is not significant. (Half power yields a reduction of 3dB, which is definitely audible.)

Consider also: Most home-type loudspeakers have a sensitivity of around 87dB (at 1 meter distance) for 1 watt of input signal. Assuming reasonable amplitude linearity in the speakers, this translates into 97 dB of output signal for 10W in, and 107dB for 100W. An input of watts would produce 109dB, which is quite loud enough to satisfy anyone but a heavy-metal junkie (footnote 3).

Perhaps, in an amplifier rated honestly, current capability is a more important determinant of maximum output level than power output into 8 ohms. Whatever the reason, the "modestly rated" SA-1s cranked out higher levels more cleanly through my speakers (Infinity RS-1Bs and MartinLogan Monoliths) than most 200-watters have. In fact, the SA-1, like the S/500 (the last Threshold amplifier I tested), might best be characterized as unflappable. No matter what program material I threw at it, it just passed it along as though sizzling cymbals and floor-shaking bass-drum thwacks were routine fare.

Sonically, this is one of the most gorgeous power amps I've heard! Highs are silky-sweet and open, with an ease and naturalness equalled only by the Jeff Rowland Model 7s (which just happen to cost about the same amount—for 200Wpc). Bass is impressively deep, tight, and orderly, maintaining firm control over even the floppiest woofers. The SA-1 is the only amp I've heard that controls the somewhat underdamped bass from the Monoliths better than the Eagle 2. Bass drum, kick drum, pipe organ, and synth bass are as well reproduced by these amps as I have heard them.

The SA-1 is ideal for driving the Monoliths, as it elicits less hardness from the speakers than does the Eagle 2, without sacrificing any of the latter's positive attributes. The SA-1 is also very well suited for driving either the high end or the low end of the Infinity RS-1Bs, though I think you'd be crazy to use the amp on the low end and discard half of the SA-1's superb performance. I tried them on the low end of the RS-1Bs because they were there; I could not justify the expense unless your checking account is so bloated that the bank is charging you to manage it.

The midrange of the SA-1 is almost perfectly neutral, being neither forward nor recessed. Yet its soundstage width and depth are excellent, with solid imaging (from good recordings) out to well beyond the placement limits of the loudspeakers. Only the Conrad Johnson Premier Fives have more apparent depth and perspective than the SA-1s, and the Fives aren't in the SA-1's league when it comes to low-end extension, control, and impact.

There is little more to be said. The Threshold SA-1 is a superb amplifier in every respect, and unquestionably one of the best that money can buy. But then, considering that it is also one of the most expensive that money can buy, it damn well oughta be good!

J. Gordon Holt wrote again about the SA-1 in April 1986 (Vol.9 No.3):

This amplifier has been in constant use in my system for over a month now, and I swear its sound continued to improve, gradually, for the first three weeks of that time.

Combining extraordinary sweetness with remarkable detail, depth, breadth, focus, and low-end impact and control, this is now, hands-down, the best-sounding power amplifier I have used.3 it has become my preferred amplifier for driving the MartinLogan Monolith speakers (displacing my now-obsolete Electron Kinetics Eagle 2), as well as just about every other speaker.—J.Gordon Holt

Footnote 1: This refers to a circuit which adjust the output-transistor bias for optimum (minimum distortion) operating conditions at all signal levels.—J.Gordon Holt

Footnote 2: A bootstrap is an emitter-follower circuit in which the base resistor is returned to a tap on the emitter resistor. This arrangement produces a "round-robin" condition in which the device hikes its own effective input impedance by using part of its through (collector-emitter) current as a signal-controlled current source feeding its own input.—J.Gordon Holt

Footnote 3: Yes, but that's figuring an 8 ohm loudspeaker; very few loudspeakers are actually an 8 ohm loud, which is why the 4 and 2 ohm power measurements of an amplifier are so significant. I think the key measurement in an amplifier is how easily it delivers a lot of power over a short time into a low impedance; even using "dynamic headroom" measurements, this figure is not readily available.—Larry Archibald

Threshold Audio
PO Box 41736
Houston, TX 77241
(713) 466-1411