Ayre AX-7 integrated amplifier Measurements
The warning in Ayre's manual is clear: the AX-7's speaker outputs should not be connected to ground or to each other.
I had almost finished my set of measurements when Murphy struck. Changing my dummy load from 4 ohms to 2 ohms, I inadvertently connected the two channels' negative terminals together. The rear-panel fuse immediately blew, but the amplifier remained resolutely dead when the connection was broken and the fuse was replaced (footnote 1). This test report is therefore missing my usual high-frequency intermodulation tests and any estimate of signal/noise ratio. But the AX-7's performance on these tests can be inferred from the tests I did perform.
The AX-7's chassis was hot after the one-hour preconditioning period at one-third power, and the metal mesh covers over the internal heatsinks were too hot to touch. The Ayre's stepped volume control offers 66 steps, the front-panel display appropriately showing bright blue numbers from "0" to "66." The unity gain setting was "33" for both balanced and unbalanced inputs. However, when I tried to measure the maximum voltage gain, I ran into stability problems with volume-control settings higher than "60" when the output was greater than about 1W into 8 ohms. The right channel started "motorboating," and the signal traces on my oscilloscope were overlaid with a low level of RF "fuzz." This happened no matter how I experimented with the grounding of my test setup, and it limited the effective gain of the AX-7 to a still-high 29dB or so.
I have experienced this before with an integrated amplifier that combines a passive front-end with a power amp of higher-than-usual gain. In practice, it should not prove a problem, as I can't conceive of modern signal sources that have such a low maximum output level that the Ayre's owner will need to use the amplifier's volume control at settings above "50." But it did alarm me.
That out of the way, the AX-7's input impedance was to specification, at 23k ohms unbalanced and 44k ohms balanced, both figures remaining constant from 20Hz to 20kHz. The amplifier preserved absolute polarity. Its output impedance was moderately high for a solid-state design, at 0.4 ohm across the audioband. As a result, the response fluctuations resulting from the Ohm's Law interaction between the amplifier's source impedance and the manner in which our simulated loudspeaker's impedance changes with frequency reached ±0.35dB (fig.1). With resistive loads, the AX-7's response was flat within the audioband, gently rolling off above 30kHz. The exact rolloff depended on the load impedance; the output was 1.25dB down at 200kHz into 8 ohms, -2.5dB at 200kHz into 4 ohms.
Fig.1 Ayre AX-7, frequency response at (from top to bottom at 2kHz): 2.83V into simulated loudspeaker load, 1W into 8 ohms, 2W into 4 ohms, 4W into 2 ohms (1dB/vertical div., right channel dashed).
As a result of this extended HF response, the AX-7's reproduction of a 10kHz squarewave was nigh-on perfect (fig.2). Channel separation was very good: better than 76dB below 20kHz, and buried beneath the amplifier's low noise floor below 1kHz. But, as can be seen from fig.3, the crosstalk does increase with increasing frequency, due to the usual capacitive coupling between the channels.
Fig.2 Ayre AX-7, small-signal 10kHz squarewave into 8 ohms.
Fig.3 Ayre AX-7, channel separation (10dB/vertical div., R-L dashed).
Even with its lack of overall loop feedback, the AX-7's distortion level was moderately low into 8 ohms (fig.4, bottom pair of traces). However, it more than doubled each time the load impedance halved, implying that low-impedance speakers are best avoided. Note that the THD remains low above the audioband, which implies that the AX-7 would have offered good performance on the HF intermodulation test. At low levels and into high impedances, the distortion was almost pure third harmonic (fig.5), which, while not quite as subjectively benign or as easily masked as the second harmonic, is still musically consonant. At higher powers and into lower impedances, the third harmonic remained the highest in level but was joined by higher-order components, as well as by some second-harmonic (fig.6).
Fig.4 Ayre AX-7, THD+N (%) vs frequency (from bottom to top at 2kHz): 2.83V into simulated loudspeaker load, 8 ohms, 4 ohms, 2 ohms.
Fig.5 Ayre AX-7, 1kHz waveform at 1W into 8 ohms (top), 0.037% THD+N; distortion and noise waveform with fundamental notched out (bottom, not to scale).
Fig.6 Ayre AX-7, spectrum of 50Hz sinewave, DC-1kHz, at 40W into 8 ohms (linear frequency scale).
While taking these measurements, I noticed something that I have seen only rarely in amplifiers: a change in steady-state test conditions led to a small, long-term drift in the state of the output. For example, to measure the percentage of distortion in its output, I set the AX-7 up to drive a 1kHz tone at 1W into 8 ohms. Suddenly changing the load impedance to 4 ohms dropped the output voltage and raised the level of harmonic distortion, as expected. What was not expected was the fact that the AX-7's voltage level into the lower impedance then began to rise again, reaching a slightly higher steady-state value after about 30 seconds. I blew the main fuse before I could further investigate this matter, but the lack of loop feedback means that the dependence of the amplifier's operating conditions on the music and load are not masked, as they are in feedback designs.
With both channels driven, the AX-7 didn't meet its specified output power at our usual 1% THD definition of clipping. Fig.7 shows that 51W were available into 8 ohms (17.1dBW) and 87W into 4 ohms (16.4dBW). The specified 60W into 8 ohms and 120W into 4 ohms were delivered at 6% and 10% THD, respectively. The shortfalls—0.7dB into 8 ohms, 1.4dB into 4 ohms—are not significant, however.
Fig.7 Ayre AX-7, distortion (%) vs 1kHz continuous output power into (from bottom to top): 8 ohms, 4 ohms.
Considering the fact that Ayre's Charles Hansen eschews negative feedback, the AX-7 still offers excellent linearity and basically good measured performance, provided the amplifier is not asked to drive speakers with an impedance that drops below 4 ohms. However, I admit to being a little alarmed by the AX-7's onset of instability at its highest volume-control settings. Assuming the review sample wasn't broken, it is fair to point out that this will almost never happen in real life.
Postscript: When he received the preliminary preprint of Stereophile's review of the AX-7, Ayre's Charles Hansen was disturbed by my finding that the amplifier was unstable at volume-control settings higher than "60." As this was not typical of the amplifier's behavior on the test bench, he suspected that the sample we'd reviewed was defective, and arranged for a different sample to be dispatched to Art Dudley.
Art reported that the second sample sounded the same as the first. He continued his auditioning with the second sample, investigating the effects of balanced operation.
When I received the second AX-7 back from Art, I hooked it up to my Audio Precision System One with the same grounding arrangement I'd used for my measurements of the first sample. After a warmup period, the new AX-7 showed none of the signs of instability at high volume-control settings that I had experienced with the first sample. Feeding this amplifier with a single-ended 1kHz tone at 100mV and setting the volume control to "61"—the conditions that had led to the first sample's visible "furring" of the waveform on the oscilloscope screen, as well as the "motorboating" (very-low-frequency changes in the output level)—produced nothing untoward. The second sample behaved impeccably on the test bench.
This suggests that there was something wrong with our first sample of the Ayre AX-7. However, as Art's experience was that the two samples sounded identical, it didn't appear to affect the amplifier's sound quality. It is also fair to note that I found no obvious reason the amplifier should sound so much better when driven by balanced sources.—John Atkinson
Footnote 1: In hindsight, this might have been because the only suitable fuse I had to hand was a 2.5A fast-blow. Ayre specifies a 4A slow-blow.—John Atkinson