Ayre Acoustics AX-5 integrated amplifier

The only thing better than a review that writes itself is a product with a compelling story. Although the latter asks a little more of us here, it's usually the more enduring pleasure.

So it goes with the new AX-5 amplifier ($9950) from Ayre Acoustics, in which designer Charles Hansen has both revived an overlooked technology from a half-century ago and brought to market a more affordable embodiment of one of his own most well-received products. All this comes in a package that requires no more room than the average electric typewriter, and that takes the place of everything you might normally put between your digital source components and your loudspeakers. It breaks a few rules—and almost breaks the dreaded five-figure price barrier—but the AX-5 is, in fact, an integrated amplifier, if one whose like you may not have seen till now.

That all-but-forgotten technology is the diamond circuit: a gateable bridge network of four bipolar transistors first described in 1964. Imagine, on the left-hand side—the signal-input side—a PNP transistor, and, below it, an NPN transistor, tied together at their bases; and, on the right-hand or signal-output side, an NPN transistor and, below that, a PNP transistor, this pair having their emitters tied together. The emitters of the left-hand transistors are tied to the bases of the adjacent right-hand transistors, and the collectors are biased all around: negatively in the case of the PNPs, positively for the NPNs. When drawn in the manner I described, with connecting nodes at north, south, east, and west, the schematic assumes a diamond shape, hence the name.

Richard Baker of MIT, to whom the patent for the diamond circuit was assigned in 1967 (he applied in 1964), described it as having a number of strengths: It can operate effectively in a floating- or above-ground condition; it can produce considerable power gain; it's reliable; it's fast; and, perhaps best of all, the diamond circuit is simple. (Baker's invention is also uniquely adaptable to modular construction, a distinction that, while not germane to this review, makes his bridge network especially well suited for use in integrated circuits, including some contemporary D-to-A converter chips.)

Charles Hansen, a self-described amateur historian of audio technology, acknowledges those advantages and adds one of his own estimation: that the diamond circuit, used as an output section, simply sounds better. The reason? Hansen suggests that, when compared with other solid-state push-pull topologies—in which two phases of a signal are recombined to form a full wave—the diamond is the only one in which the two half-signals are joined at a single point in the circuit, with no intervening circuitry. Thus, the diamond circuit creates an output that's more faithful to the shape of the input (footnote 1). Ayre biases the diamond output section in the AX-5 to operate in class-A/B, whereby it delivers 125Wpc into 8 ohms or 250Wpc into 4 ohms.

Hansen's other remarkable idea for the AX-5 is a circuit innovation called variable-gain transconductance or VGT, first seen in Ayre's top-of-the-line KX-R preamplifier of 2008 ($18,500). As Hansen explains, most active preamplifiers work by applying to the input signal a certain amount of voltage gain, so the signal can effectively drive a power amplifier. But in order for there to be a reasonable volume range—and to simply keep the playback level from being too loud—the voltage-gain stage is preceded by a potentiometer, which attenuates the signal. The drawback of this is that such a preamp will exhibit its maximal signal/noise ratio only at its maximal (unattenuated) volume. As Hansen puts it, "Since most preamps are used anywhere between –10dB and –40dB for an average listening level, this means the S/N ratio in actual use will be 10–40dB worse than on the spec sheet."

As implemented in the AX-5—which doesn't incorporate a preamplifier stage per se—Ayre's VGT circuit allows the user to determine how much gain is generated by the amplifier's input stage, which itself comprises a total of four complementary-differential JFETs. The volume knob on the AX-5's front panel controls a pair of enormous, motor-driven, Shallco silver-contact rotary switches, each of which contains dozens of hand-selected, low-noise resistors. Every volume-level adjustment made by the user has the effect of switching into the AX-5's input-stage circuit a different set of resistors, the values of which alter the transconductance of those JFETs—and thus calls into play a specific level of gain corresponding with that setting. The volume system has 46 steps of 1.5dB each, over a range of 67.5dB. (I'm told that, by changing a single resistor in each of the AX-5's channels, one can adjust the overall gain range to accommodate, say, speakers that are significantly more or less sensitive than average.) Thus the AX-5 doesn't use signal attenuation at all, but rather creates variable input-circuit gain, on demand, to suit the desired volume level.

In keeping with Ayre's long-standing practice, the AX-5 is not only a zero-feedback design; its circuitry is fully balanced from input to output. Interestingly, Charles Hansen endorses balanced technology not for its ability to reject hum and noise that might enter the circuit through cables, but because it (similarly) rejects the hum and noise that could enter the signal path from the power supply. "When the day comes that a totally perfect power supply is developed, there will be no more need for using balanced circuitry in home-use applications," he has stated.

The circuitry for the AX-5's fully analog power supply is located near the front of the chassis: ie, as far as possible from the input circuitry, much of which is snugged against the rear panel. Input selection is accomplished with FET switches, implemented with Ayre's own supporting circuitry, a technology the company has employed for a number of years, and which Hansen praises as being superior to all others in noiselessness and reliability. Between the left- and right-channel input boards one finds the big Shallco rotary controls, fronted with a very high-torque Lin step motor and the custom mechanicals used to work the switches: an impressive dual-mono system, synchronized with a toothed polymer belt. A fairly enormous EI-frame power transformer occupies the very center of the chassis, straddled by the left- and right-channel output sections and their generously sized heatsinks.

Ayre says they designed the AX-5 so that its dimensions would correspond with the infamous golden ratio, in common with the other 5 series in their line. I'm baffled as to how it qualifies for such a distinction—according to my ruler, and ignoring jacks, knobs, and other protrusions, the AX-5's case is almost perfectly square. That said, I was impressed by an observation from Ayre's Alex Brinkman, who says that, after their ideal dimensions were selected, "we got the transformer we wanted, with the specs we wanted—and it was too big to fit in that chassis. So rather than make it even bigger, we cut an opening in the bottom of the chassis, to accommodate the transformer's height." Indeed they did, and rather neatly, I'd say. The finished box is made of aluminum, with a brushed exterior. Most of the fasteners are stainless steel, and the quality of fit between various panels was all right; but the panel edges were sharper, and their overall feel less smooth, than I would wish to see in a product that sells for nearly five figures.

Installation and Setup, Part One
The AX-5 comes equipped with four balanced inputs (XLR jacks) and two single-ended inputs (RCA jacks): Charles Hansen reasons that it's easier and less expensive to adapt a balanced input for single-ended use (footnote 2) than the other way around. (And the other way around will always be somewhat compromised.) In any event, given that fully balanced operation is one of Ayre's calling cards, it's reasonable to expect that the prospective buyer of an AX-R would not be troubled by the two-to-one ratio in favor of balanced inputs. Indeed, even though I don't keep a balanced system—which, I gather, is only slightly less challenging than keeping a kosher kitchen—I spent slightly more time using the amp with a borrowed sample of Ayre's balanced QB-9 D/A converter than my own single-ended Sony SCD-777ES SACD/CD player.

Prior to use, either the buyer or the buyer's dealer must access the AX-5's Setup mode in order to configure and activate one or more of its six inputs, none of which is usable straight out of the box. In Setup mode, both the input-selector and volume knobs—on the left and right sides of the front panel, respectively—are used to scroll through and select various bits of information, typically with the aid of two front-mounted button switches. Configuration didn't take terribly long—it's a software thing, done from outside the amp, no tools required—and would seem to be within the capabilities of anyone with the patience to read the manual and to follow the steps as written. (None of the AX-5's setup regimens was sufficiently intuitive that I could perform it without manual in hand, even after weeks of use.)

Once configured, the AX-5 was simple to use. With AC power applied by means of a hefty rear-panel rocker switch, the Ayre is roused from its low-power-consumption mode—in which bias current is removed from the collectors of the output transistors—with a push of the right-hand button. (Subsequent to that, a brief press of the same switch effectively mutes all inputs; a long press returns the AX-5 to low-power mode.) From there, things are straightforward: Use the left-hand knob to select the desired input (only inputs that have been configured will show up as choices: a boon, I think, for those of us who dislike the cognitive clutter of choices in which we have no interest), the right-hand knob to adjust the volume. As to the latter, the electromechanical noise produced by each step up or down the ladder of loud takes some getting used to—sounding, as it does, like a roof leak dripping into an empty metal pail. After a day or two, I found it rather charming.

Footnote 1: This, if you don't mind my saying, teeters on the verge of saying that the diamond output is closer than other push-pull circuits to the single-ended ideal. You are free to imagine a smiley emoticon at the end of that sentence.

Footnote 2: Ayre makes and sells such an adapter ($75/pair).

Ayre Acoustics, Inc.
2300-B Central Avenue
Boulder, CO 80301
(303) 442-7300

tmsorosk's picture

Great review art .

Et Quelle's picture

The AX-5 is too rich for my blood. Shipping is undoubtedly free though any
extras would put you over the 5 figure price barrier.
The diamond cut circle is simple but not the best things. Superficial looks
mean the most in audio only. A 46 step volume system and Shallco switches, I guess justify Ayre going higher than most of their other products. I sure hope Stereophile shows an opened pic of future amp and it has a huge transformer in it like that. I have a tube preamp and solid state amp on my wish list but meanwhile I will use my Nano phono preamp.
Great job on the configuration and the review. You wouldn't want to spend too long on something like that!

audiofrk's picture

So Art can I ask from what you remember from the AX-7 is the AX-5 a step up (same audio quality but more power) or a new league (sounds better than the AX-7)?