Commentary on the audiosciencereview of the Ayre Codex DAC/Headphone amplifier

A review of the Ayre Codex by Audiosciencereview

I bought an Ayre Codex for myself last year to find out how my own diy efforts, the Sapphire headphone amplifier, would compare to a serious high end product from a serious high end company. I had read the audiosciencereview report above before I decided to buy it, as well as many other online reviews. It did not dissuade me because I had faith in Ayre and their design philosophy. This is only reasonable: the Sapphire shares much of the same design principles of the Codex, and I know from experience that the measurements we make to test audio equipment rarely correlate with perceived sound quality. That the Codex posted some disturbing test scores did not bother me. Put it another way, I've had plenty of gear pass through here that won great praise for measurements but sounded like cr*p...

In the next few posts I'll be providing some annotations to the Codex review based on my own listening experiences, cross-referenced with my thoughts about how the Sapphire sounds in comparison. Expect some soapboxing about Ayre and the commercial hifi industry in general, together with come color commentary on the measurements-vs-listening-tests debate.

Stay tuned!

 

[rjm]

Please review this excellent thread on the Codex over at head-fi. There are plenty of photos and it's interesting to read the expectations and opinions expressed by folks at the time the product was first announced.

The picture of the internals reproduced below clearly indicates the conflict arising from wanting a small, desk-friendly enclosure and wanting a fully discrete analog headphone amplifier. Two EI laminate power transformers placed dead center of the signal layout is far from ideal. While the circuit board seems otherwise nicely designed, the lack of distributed capacitance stands out. There are six large filter electrolytics (the green cans) as part of the power supply, but no smaller electrolytics in bypass roles on the lower half of the board where the amplifier stages are located.

 

[rjm]

To set the stage, let me introduce the distortion measurements for the Ayre Codex. This is going to be central to the discussion later, so we might as well get it out of the way. Audiosciencereview measured the 3rd harmonic at about -68 dB at 1kHz. Stereophile measurements by John Atkinson recorded about -60 dB 3rd harmonic at 50 Hz (see Fig 7). Note that in both cases the output is essentially unloaded, the posted numbers characteristic of the Codex's amplifier circuitry rather than load-dependent.

Audioscience review - "Those THD+N numbers are almost two orders of magnitude worse than what I expected to see! SINAD of 67 to 68 dB is unbelievably low for a non-tube product."
John Atkinson - "As I have come to expect from the Ayre Acoustics design team, the Codex offers excellent measured performance."

Leaving aside the why, the "how" of it is understood when you find out how hot the Codex runs idle. The external chassis is cooking. (Side note: none of the components inside are heatsinked as far as I can tell from the photos.) This high operating temperature indicates to me the output stage is running very high bias currents. Doing so increases the rated maximum class A output power output but also increases distortion, regardless of whether any headphones are connected or not. This could be addressed by adding feedback of course, but Ayre chose not to. It could also be addressed by reducing the bias current. Again, Ayre chose not to.

 

[rjm]

Because - spoiler alert - distortion colors the sound and 3rd harmonics running near 0.05% (-68dB) is significant.
By comparison, my Sapphire4 as presently configured in closed loop mode posts 0.001% total harmonic distortion (mostly 2nd harmonic) and -93.9 dB THD+N.

Distortion measurements just provide a snapshot of the performance at a single signal amplitude and frequency, however. It's an excellent way to check for design flaws, but I wouldn't hold it as an absolute indicator of sound quality. An op-amp headphone IC will show lower distortion than the Sapphire4, for example (<0.0005% THD) but you'll never convince me it sounds as good. The Codex is in a grey area where the distortion numbers are high enough to flag as a potential flaw but not high enough to suggest an actual fault.

 

[rjm]

I evaluated the Codex using Sennheiser HD-600 headphones with balanced cables. The Codex was connected by USB to my Surface laptop running Windows 10 and using the Microsoft native audio drivers, all processing off. The Codex has no analog input to test the output stage separately from the built-in DAC, but I have no reason to think the DAC and digital volume control are anything other than top-notch.

The Codex is enjoyable to listen to. The overall timbre is neutral, detailed, and appealing. The soundstage is coherent and natural. Rhythmically it all sings nicely. It has two weak points: Lack of ambient clarity and lack of dynamic grip. The former can be described as muddiness/fuzziness. The latter, anemic, bloomy bass frequencies. With the Codex, my HD-600s (300 ohm load and admittedly a bit of a tough cookie to drive) sound distinctly soft. Neither weakness makes the Codex sound bad, not at all. What I'm hearing here is the output from a truly wonderful DAC/digital volume control being relayed through a highly competent but not outstanding headphone amplifier.

 

[rjm]

Ok, let's try to wrap this up. I've spent a lot of time comparing the Codex (with its built-in DAC) against the Sapphire 4 + ASUS Xonar STX DAC combo. Sennheiser HD-600 headphones. As a side note I compared balanced and single ended connections from the headphones to the Codex and found the balanced setup to sound considerably cleaner with a more open soundstage.

I preferred the Sapphire 4. No instance that I could find where I thought the Codex sounded better. The Codex fails to energize the HD-600s, the bass frequencies lack impact and drive. Rhythm and dynamics suffer. Comparatively speaking, music is less interesting to listen to. The only other point I could reliably detect is a slight thickening of the harmonics. This can be described as added tonal complexity and warmth, or as a lack of tonal purity. Both takes are valid in my opinion. It's not an unpleasant addition, but it does skew the musical picture somewhat.

Just how different music sounded through the two systems surprised me, to be honest. There is not, objectively, much difference in quality yet the change in presentation, in terms of musical focus, is dramatic.

My final word on this is the Codex is a nice-sounding headphone amp that you should listen to carefully before buying. It may work with your headphones better than it did with mine. Don't pay the distortion measurements much concern. Distortion of a 0 dB (1 V rms) test signal is only loosely correlated to the distortion you'll experience during normal listening. This is doubly true for a zero-feedback design. While I could convince myself that I could hear the higher distortion relative to my reference this was reaching the limits of my detection ability -- and anyway unrelated to my main gripe with the system, which was the missing low-frequency drive. As for the question of whether there is any measurement you could perform that would reliably quantify the difference in bass quality I heard, I highly doubt it. Nothing here changes my previous conclusions on this topic: In audio, measurements are useful until they aren't. Measurements will tell you if you did something stupid and help you figure out what the problem is. Good sound derives from rigorous design principles refined by ear.

 

[rjm]

A Coda to the Codex: I remembered somewhat late in the game that the Codex also has line outputs, both balanced and single-ended. While (as far as I know) the line output is taken from the same amplifier stage as the headphone output (and should therefore sound the same) there could be a difference as a result of the very different loading (300 ohm headphones vs. 20k or more for a preamp, headphone amp etc.).

So I tried it in DAC mode, connected to the Teac HA-501 via XLR.

I was pleasantly surprised. It was still a little bit "fat", but far cleaner than when driving headphones directly. There was a spaciousness and warmth ("musicality" if you will) present that I had been sorely missing after living with the 2Qute for several months. Enough promise to spur me to take action, and on the spur of the moment went and bought an Ayre QB-9 Twenty on a Japanese auction site, thus negating the need for both the 2Qute and Codex.

I think there's a lot to like about the Codex, and the feature set (headphone amp, digital preamp, DAC) for the price makes it a great deal. I just don't think it matches well with my HD-600 headphones. The gain is way too low, for a start. On the 0-100 scale, I have to have it near 80 to even get remotely loud. Even running balanced the HD-600s sound weak and bloated. Considerably worse than through the HA-501. With a different impedance/sensitivity that could be a very different story. Best to try before you buy.