Boulder 500AE power amplifier

To high-end audiophiles, the Boulder 500 amplifier and its less expensive derivative, the 500AE (Audiophile Edition), would not seem to be "high-end" designs. They are designed around op-amps (felt by many to be generally poor-sounding), they have scads of negative feedback (which is perhaps why op-amps sound bad), and they have only a moderately hefty power supply. Why, then, is Stereophile publishing a review of an op-amp–based power amplifier? Read on...

The op-amp (for operational amplifier) circuit module was originally intended to perform simple mathematical operations in all-tube analog computers. (Does that date it, or does it?) It combines several active devices to produce a DC-amplifying package having, in its "perfect" form, infinitely high input impedance, infinitely high gain at DC, and zero output impedance. A simplified solid-state op-amp circuit is shown in fig.1. The input is applied between the bases of the two input transistors (Q1 and Q2), so that it appears in opposite phase at their collector outputs. The output of one input transistor (Q1) then goes through a gain stage (Q3), which inverts its polarity to match that coming out of Q2. Both signals are then combined, amplified again by Q4, and fed to a complementary-symmetry emitter-follower output consisting of Q5 and Q6.

Fig.1 Simplified circuit of an op-amp.

Note that, as befits a DC amplifier, there are no capacitors in the signal path—which is always to the good in audio, because even the best caps (don't ask me what they are) degrade the sound to some extent. In addition, the op-amp's differential inputs provide an inherently balanced input, as well as allowing the device to be used inverting or non-inverting, depending on its input connection polarity and feedback connections.

Long ago—in the late 1960s/early '70s—op-amps were realized as cheap, ultracompact integrated circuits, which made them very attractive for use in audio products where competitive cost was important enough to outweigh the op-amp's shortcomings. (These then consisted mainly of poor and often asymmetrical slew rate and low output current-sourcing capability.) While a typical "raw" op-amp IC circuit has an open-loop (without feedback) gain of around 30,000 (90dB)—some have as much as 120dB (an incredible 1 million)—the open-loop frequency response is typically from DC to just 30Hz. Both characteristics are clearly inappropriate for any audio application—30Hz is half an octave below the lowest note of the double bass—but both are remediable through the use of very large amounts of inverse feedback—if you can get away with it. And "getting away with it" proved impossible with many of the more common IC op-amps—until the late Deane Jensen, best-remembered for his audio transformer designs, unveiled his JE-990 discrete op-amp circuit (footnote 1).

"Discrete op-amp?" That may sound like an oxymoron, like "business ethics," but it isn't. Op-amps were, of course, discrete before they evolved into ICs, and the discrete transistors of today are much faster than they were back in the '60s. Today, discrete op-amp design can offer lower noise, higher stability, higher current output (for lower distortion into low-impedance loads) at lower temperatures, higher output voltage, and a higher slew rate than IC op-amps. The Jensen JE-990 design that serves as the basis of the Boulder 500AE, for example, has a slew rate of around 17V/µs.

There are two JE-990s per channel in the Boulder 500AE, one serving as the input section, the other as the voltage driver and output section. Both drive a so-called complementary-symmetry pair of output devices—see Sidebar—with the main difference being that the second "pair" are two banks (per channel) of power transistors.

Because complementary-symmetry solid-state amps do not use output transformers, they have far less HF phase shift than tube amps, which means they can tolerate much more negative feedback without becoming unstable. Typically, 20dB of overall loop feedback is all a tube amp can take before it starts to ring audibly; imagine this limitation imposed on an op-amp, with an open-loop gain of more than 90dB! The Boulder 500AE has about 100dB of feedback, but none of it is overall loop feedback; it's all local feedback, involving only one or two stages at a time, which reduces even further the possibility of instability or TID (Transient Intermodulation Distortion, see Sidebar).

The product
The 500AE is solidly built, with a fold-reinforced 1/8" aluminum chassis and more than adequate anchoring of all heavy internal parts. (It looks as if it would travel well.) It has no handles, but at only 55 lbs, even frail old me had no trouble schlepping it around handleless. Gold-plated balanced inputs are provided. It's possible to drive the amplifier in single-ended mode (unbalanced) by using the optional ABL adapters.

Although hardly the arc welder of a 100 lb Krell, the 500AE's power supply isn't exactly a featherweight. With a 6"-diameter, 3"-thick toroidal power transformer and 74,000µF of storage capacitance per channel, the amp has a claimed peak current capability of 50 amperes, and is designed to be able to deliver that current for up to a tenth of a second before supply depletion starts to set in.

Because the 500AE is capable of passing DC, it is necessary to provide some sort of protection against inadvertent DC offsets that could fry the woofers. This is done in two ways. The first line of defense is a servo circuit which zeros-out small amounts of DC offset at the input, up to a point. Beyond that point, the amplifier shuts down if DC at the output exceeds 3V (1.125W into 8 ohms) for more than 0.1s. After the source of the offset is removed, recovery is spontaneous—usually in a few seconds, although if the offset was very severe, it may take minutes for recovery. In this case, it's best just to turn the amp(s) off, then on again. (Note that the DC-offset sensor is not in the signal path; it's a sidechain circuit, which has no effect whatsoever on the audio signal until the triggering threshold is exceeded.)

Strapping for mono operation is a simple matter of flipping a rear-panel switch and connecting the speaker across a pair of the Hot terminals. Incidentally, there are two pairs of output connections, each pair in parallel, presumably to facilitate bi-wiring of the loudspeakers.

Test conditions
Equipment used for this review included the Proceed PCD-2 CD player, Revox A-77 15ips 2-track tape recorder, a Sony PCM-F1/SL-2000 digital recording system, Pioneer LD-S2 laserdisc player (the state of the art!), Threshold FET-10L line controller, and the Stereophile-owned pair of Sound-Lab A-3 full-range electrostatic loudspeakers (updated to the latest version and next in line for review). Audio interconnects were Monster M-1000s, and loudspeaker cables were AudioQuest Greens. Program material ran the gamut, although most of the CDs used were orchestral, from Delos and Sheffield.

Footnote 1: See "JE-990 Discrete Operational Amplifier," the Journal of the Audio Engineering Society, Vol.28 No.1–2, January/February 1980, p.26. Website:—John Atkinson