Solid State Power Amp Reviews

Design
There's more to the SA/12e than just quality parts and fortress-like build quality. The design thoroughness begins at turn-on, in which internal components are protected against surges by a controlled power-supply charge-up. To give you an idea of the initial current draw of these amplifiers, turning one of them on caused the line voltage to dip from 118 to 106V for an instant; with both amplifiers idling, the line voltage stabilized at about 114V—a 4V drop from the level sans Thresholds. The SA/12e, being fully class-A from input through to and including output stage, draws its maximum power from the line at idle—about 1000W for the pair. Its heatsinks, however, are efficient; they get quite warm, but never to the point where you cannot keep your hand on them indefinitely. The heatsinks of the class-A Levinson No.20.5, on the other hand, operate decidedly hot to the touch despite that amplifier's considerably lower power rating.

To maintain constant bias within the amplifier under dynamic conditions, Threshold uses a patented optical bias system. By the use of so-called "opto-isolators," the necessary current information is obtained from the output stage without the bias circuitry interfering with its operation. Optically coupled circuitry also adjusts the front-end bias for perfectly balanced operation. Threshold also claims that the optical bias system maintains the active elements at a much more constant temperature than conventional bias circuits, reducing the stress on these devices and thereby prolonging their life.

The input of the Threshold uses proprietary circuitry requiring no additional active components for balanced operation beyond that used in the unbalanced mode. The input stage itself consists of N-channel J-FETs in a cascoded configuration having a high input impedance and only a few dB of local feedback. In fact, one of the design criteria of the entire line of Threshold amps appears to have been the elimination of overall, "global" feedback (footnote 4).

Threshold is not the only high-end company attempting to minimize or eliminate global feedback, and not without some justification. In theory, feedback (footnote 5) gets you "something for almost nothing." It trades gain (which modern solid-state active devices can provide in cheap abundance) for improved linearity (lowered distortion) and improved bandwidth. But the case against negative feedback argues that it really only performs its function perfectly with continuous signals. It thereby makes another tradeoff: good test-bench figures for less predictable performance with real-world, noncontinuous musical signals. It must be said here that not all high-end amplifier manufacturers agree with this; some have produced well-received amplifiers, using significant amounts of feedback, to prove their case. Still, the prevailing trend appears to be the minimization of global negative feedback.

With or without feedback, the basic problem in any amplifier is how to get its output to exactly replicate its input, only more so. The easily understood concept of anemic little electrons struggling from the preamp to the input of our power amplifier, where they enter a Gold's Gym of circuitry, only to be pumped up to do battle with our awaiting loudspeakers, is serviceable but misleading. In actuality, the inputs of our amplifiers merely act to control the reservoir of current capability in the power supply. The varying voltage of the input is increased by the early stages of the amplifier. Finally, the output of the last voltage amplifier stage literally acts as a gate on the output transistors, which provide power to the loudspeakers sourced from the amplifier's power supply in a manner which varies in accordance with this control voltage. At least that's how it usually works.

Threshold's approach to providing the required drive to the loudspeakers without the need for global feedback is slightly different, and is centered around their patented STASIS circuitry. This configuration, developed for an earlier generation of Threshold amplifiers, is continued in the new "e" series. A small, high-quality, class-A-operated voltage amplifier is connected directly to the load (the loudspeaker). This amplifier must be able to deliver, in theory, whatever voltage is required across the load to provide the sound level demanded—within the power limitations of the amplifier. Notice that I said "in theory." In actuality, it would not be able to do this unaided; a voltage amplification stage does not like to supply current into low impedances of the sort typical of our loudspeakers. The output voltage sags badly under load (which is, simplistically, why your preamp won't make much of a sound driving your loudspeakers directly).

Threshold's voltage amplification stage is a bit more capable than that: it will put out between 5 and 10 amps of current before its output voltage linearity starts to suffer. To provide the remainder, a current source (which Threshold refers to as a current mirror bootstrap) is connected both to the load and to the voltage amplifier. The latter then establishes the required voltage while the former (actually the output transistor complement of the amplifier) furnishes the current needed through the load—drawing on the power supply to provide it. The voltage amplifier is therefore able to maintain the required voltage without sagging. (Of the 64 output devices in the circuit, 60 appear to be in the "current mirror bootstrap," four in the voltage amplifier stage.) Put more simply, in the STASIS design the final voltage stage and the output (power) stage operate in a symbiotic, tandem arrangement. The bottom line: high intrinsic linearity without recourse to global feedback.

All Threshold amplifiers incorporate this STASIS circuitry. The differences from one amplifier to the next are in the number of output devices and the size of the power supply—altering the available power—and in the use of full class-A bias (the SA series) or a combination of class-A and class-AB bias (the S series, which operates at class-A until about 20% of rated power, then transitions to -AB) in the final current bootstrap section. In the case of the class-A amplifiers, of which the SA/12e is the most powerful, they will operate, at 8 ohms, in pure class-A at their rated power, but things become more complex at lower load impedances. The amplifiers will roughly double in continuous power output capability for each halving of impedance (to a point), but will not continue to operate in class-A up to that power level.

In the case of the SA/12e, it will put out 500W continuously into 4 ohms, but will leave class-A at 125W into that impedance. It will deliver about 800W into 2 ohms (though distortion rises to about 1% there), but only about 65W of this will be class-A. Into 1 ohm the respective values are 1300W (1% distortion), overall about 30W class-A. The SA/12e is rated to operate into impedances below 1 ohm and into highly reactive loads, but continuous operation at these extremely low impedances will be limited by the power-supply fuses and thermal limiting. Again, the bottom line: Threshold's claims would seem to argue that the SA/12e will be unflustered by any known loudspeaker load.

It should also be pointed out that Threshold rates the SA/12e for 60 amperes continuous output. I'll admit that I have a problem with this figure—it seems hypothetical at best. Sixty amperes into 8 ohms gives a continuous output figure of 28,800W, a patently nonsensical figure which bears no relationship to either the continuous power output rating or the 250W continuous dissipation rating of each output device. I asked Threshold's designer Nelson Pass about this, and he stated that this rating refers to the amplifier's ability to supply current into very low impedances—fractions of an ohm. Unless the conditions are more carefully specified, I consider the continuous power rating per se to be academic in real-world use at best, potentially misleading at worst. But Threshold is not the only company quoting output current capability.

---

Footnote 4: Feedback from the output of an amplifier to the input. As contrasted with local feedback, encompassing usually one gain-stage.

Footnote 5: I trust readers will realize that I am talking here about negative feedback. Typically, positive feedback gives you an oscillator.