Digital Processor Reviews

Ayre Acoustics Codex D/A headphone amplifier Measurements

Sidebar 3: Measurements

I measured the Ayre Acoustics Codex with my Audio Precision SYS2722 system (see the January 2008 "As We See It"). Source materials were WAV and AIFF test-tone files, played on my MacBook Pro running on battery power with Pure Music 3.0. (I couldn't get the TosLink optical input to work with S/PDIF data sourced from an Astell&Kern AK100, footnote 1) Apple's USB Prober utility identified the Codex as "Ayre USB Interface" from "Ayre Acoustics," and the serial number was given as "Streamlength(tm)." The Codex operated in the optimal isochronous asynchronous USB mode, and Apple's AudioMIDI utility revealed that it accepted 24-bit integer data sampled at all rates from 44.1 to 384kHz.

The volume control's maximum setting was "100," and it operated in accurate 1dB steps. The maximum output level at 1kHz in preamplifier mode was 7.14V from the balanced 3.5mm jacks on the front panel and the balanced line-output jacks on the rear panel, and 3.545V from the front-panel ¼" headphone jack and the unbalanced, rear-panel RCA jacks. Switching the Codex on set the volume control to "66," which was equivalent to –34dB with respect to the levels listed above. The maximum output levels in DAC mode were 4V (balanced XLRs) and 2V (unbalanced RCAs). The output impedance was 155 ohms from the balanced XLR jacks, 78 ohms from the unbalanced RCA jacks, 6 ohms from the balanced 3.5mm headphone jacks, and 3 ohms from the ¼" headphone jack, all figures constant across the audioband. (The balanced impedances are twice the unbalanced values, because there are now two output stages in series.) All four sets of outputs preserved absolute polarity (ie, were non-inverting).

All the following measurements were taken from the balanced headphone output. The Codex's impulse response (fig.1) was typical of Ayre's minimum-phase filter, with minimal ringing after the full-scale sample. The red and magenta traces in fig.1, taken with 44.1kHz-sampled white noise, show that this filter has a slower-than-usual rolloff above half the sample rate. As a result, the aliasing image at 25kHz of a full-scale tone at 19.1kHz (blue and cyan traces) is suppressed by just 10dB, and some higher-order images appear lower in the audioband, albeit at a very low level. Given the zero-feedback topology of the Codex's analog circuit, the distortion harmonics of the 19.1kHz tone are also commendably low, at levels close to –90dB. Fig.2 shows the Ayre's frequency response at sample rates ranging from 44.1 to 384kHz. As expected from fig.1, the rolloff starts earlier than usual, the 44.1kHz response (left-hand blue and red traces) down by 3dB at 20kHz.

Fig.1 Ayre Acoustics Codex, impulse response (one sample at 0dBFS, 44.1kHz sampling, 4ms time window).

Fig.2 Ayre Acoustics Codex, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at 0dBFS (left blue, right cyan), with data sampled at 44.1kHz (20dB/vertical div.).

Channel separation was superb, at >120dB in both directions below 6kHz and still >110dB at 20kHz. Spectral analysis of the amplifier's low-frequency noise floor (fig.3) revealed a very low level of random noise components, and while some spuriae can be seen at 60Hz and its odd-order harmonics, presumably due to magnetic interference from the power-supply transformer, these are also very low in level. The low level of analog noise means that the Codex offers close to 19-bit resolution, increasing the bit depth of a dithered 1kHz tone at –90dBFS from 16 to 24 dropped the noise floor by 16dB (fig.4). However, while the waveform of an undithered tone at exactly –90.31dBFS was well presented, it was overlaid with some high-frequency noise (fig.5).

Fig.3 Ayre Acoustics Codex, frequency response at –12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel blue, right red), 96kHz (left green, right gray), 192kHz (left cyan, right magenta), 384kHz (left blue, right red) (1dB/vertical div.).

Fig.4 Ayre Acoustics Codex, 24-bit data, spectrum of 1kHz sinewave, DC–1kHz, at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).

Fig.5 Ayre Acoustics Codex, spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with: 16-bit data (left channel cyan, right magenta), 24-bit data (left blue, right red) (20dB/vertical div.).

At maximum level, a low-frequency tone was accompanied by primarily the third harmonic, at –60dB (0.1%), though some second harmonic was present, particularly in the left channel (fig.6). Into lower impedances, the third harmonic dropped a little but the second and fifth rose slightly, to –75dB. Tested with an equal mix of tones at 19 and 20kHz with the combined waveform peaking at 0dBFS, intermodulation products were actually low in level, but the spectrum in fig.7 is dominated by aliased spuriae.

Fig.6 Ayre Acoustics Codex, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data (left channel blue, right red).

Fig.7 Ayre Acoustics Codex, 16-bit data, spectrum of 50Hz sinewave, DC–1kHz, at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).

Finally, although the Dunn-Miller J-Test tone is not diagnostic for USB digital data, in which the clock is not embedded, the Ayre Codex performed well with both 16-bit data (fig.8) and 24-bit data (fig.9). There was just a very slight accentuation of the sideband pair closest to the 11.025kHz tone with 16-bit data, and while some supply-related sidebands can be seen in both graphs, these are extremely low in level.

Fig.8 Ayre Acoustics Codex, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms, 44.1kHz data (left channel blue, right red; linear frequency scale).

Fig.9 Ayre Acoustics Codex, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit USB data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

Fig.10 Ayre Acoustics Codex, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 24-bit USB data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

As I have come to expect from the Ayre Acoustics design team, the Codex offers excellent measured performance.—John Atkinson


Footnote 1: The Codex manual does warn that the optical input will not be compatible with all devices, and that there is a firmware fix.
COMPANY INFO
Ayre Acoustics, Inc.
2300-B Central Avenue
Boulder, CO 80301
(303) 442-7300
www.ayre.com
ARTICLE CONTENTS

COMMENTS
headphile's picture
Submitted by headphile on May 25, 2016 - 9:49pm

It seems like a missed opportunity to not have Tyll Hertsens at least co-author headphone reviews when they are posted to Stereophile instead of Innerfidelity (Stereophile's review of the Chord Hugo TT is another example of this). I am an avid reader of The Enthusiast Network's audiophile websites and I have consistently found Tyll to be a trusted reviewer with invaluable headphone experience. His reviews include accurate comparisons and measurements that give useful context that can be used to make informed purchase decisions. I hope in the future more reviews like this will at least be co-authored by Tyll or posted to Innerfidelity.

Jon Iverson's picture
Submitted by Jon Iverson on May 26, 2016 - 10:41am
Great idea - will see if this is possible for the next headphone/DAC review.
cgh's picture
Submitted by cgh on May 26, 2016 - 10:26am

... watch stuff grow.

lennykp's picture
Submitted by lennykp on May 26, 2016 - 9:52pm

Did you try AYRE Codex as balance preamp? how does it perform?
Please ask manufacturer if USA model work on 240V also?
Cheers

acuvox's picture
Submitted by acuvox on October 31, 2018 - 8:29am

Ayre Acoustics has always scrupulously minimized or avoided generating RFI signals in its boxes. This means putting the control micro to sleep or hibernating, analog power supplies and static, non-matrixed displays.

This example is retro visually. The seven segment LED display goes back to the first pocket calculators from Bowmar in 1971, and the first digital watch in 1967. Pin count was a major issue, so calculator displays where multiplexed, switching the digits on and off in succession on the first microprocessors.

For only two digits, you just need 14 pins for the segments to run on DC. The extra 5 pins are insignificantly expensive in parts cost and circuit board space, but this critical design decision improves sound quality better than expensive internal mu metal shielding.

X