This is a review and detailed measurements of the Ayre CODEX balanced DAC and headphone amplifier. It is on kind loan from a member. The CODEX is a current product but seems like it was released 3 or so years ago. It costs USD $1,795.
I am puzzled by the look of the CODEX as it doesn't match the rest of the Ayre products:
If you are not familiar with Ayre, it is a US company and is probably best known by its late founder, Charles Hansen. He has been active on forums for years. The company is in high-end audio business so my expectations were high on this DAC and headphone amplifier.
I was pleased to see a set of balanced XLR connectors on the unit and inclusion of mains power supply despite its rather diminutive size:
Then again, closeness of the mains input gives me some pause regarding its noise immunity for the RCA outputs.
In use the CODEX runs very warm. The marketing material talks about push-pull, discrete power supply which likely is the cause (it is a little power amplifier that works on DC).
I did all of my testing with USB as usual. The CODEX was plug-and-play with Windows 10 Creators edition.
Audio Measurements
As usual I start with the same dashboard profile (test settings) for all DACs. Here is the performance with RCA output:
The unit is in "DAC mode" meaning volume control is out of the loop. I was disappointed to see the output voltage not reaching nominal 2 volts. But the biggest disappointment was in the distortion department. 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. I thought something may be broken here so I ran the same test with balanced XLR output:
Mains leakage is reduced but that is not our problem. Our problem are those massive harmonic distortion spike with the 3rd harmonic at 3 kHz reaching way up to -68 dB. As such, by itself sets the SINAD.
I could not believe my eyes. I was ready to take out another DAC to make sure the analyzer was not broken. Then I read some write-ups and figured out the issue is with the design:
Ah, tell me it isn't so. Another believer in voodoo audio and imagined problems with feedback. Electronic parts are not linear. While we try to operate them in the most linear version, they do not have a 1:1 output with respect to input. Feedback was invented almost a century ago to deal with this problem. A small amount of the output is sampled and compared to the input. If they don't agree, the amplifier changes its gain to make them so. This happens extremely fast. There is no time delay as lay people imagine (in audio context). Without feedback I dare say you wouldn't be seeing these characters on your screen! Or make phone calls. Or stream audio/video. It is fundamental to good engineering.
These imagined improvements without feedback are just that: imagination. Anyway, let's continue testing.
The SINAD figure naturally lands the CODEX near the bottom of the pile of DACs out of some 125 tested so far:
Right there with Audio-gd which likewise schews negative feedback.
Jitter performance is good although not as perfect as I like to see in this price range:
Linearity test showed the clear benefit of balanced output and its rejection of mains and its harmonics:
With RCA output (in blue) there is a spike at 180 Hz (third harmonic of 60 Hz in US). My test frequency is 200 Hz and the filtering I use for Linearity test does not exclude that. This means that at low amplitudes (all the way to the left), the mains harmonic has significant amplitude and pollutes the final output. A larger case with better separation from RCA outputs would have helped this possibly. For now, use balanced XLR outputs.
For people who say single tones don't represent music, here is our 32-tone test signal getting subjected to CODEX's non-linearities:
As expected, we see a ton of "grass" in the skirts of our main tones instead of silence. These conspire to hide low level detail in music and cause exaggeration of high frequencies in music, making it more harsh. How anyone can argue this adds more "rhythm" is beyond me.
Intermodulation distortion shows how lack of feedback causes rise in distortion as we ask the DAC to produce higher and higher voltages:
And here is the dynamic range:
Strange to see worse dynamic range with balanced output and such a large drop.
Headphone Amplifier Tests
There are both unbalanced (1/4 inch) headphone jack and "balanced" dual 3.5 mm. I have three other configurations in my balanced test fixture but don't have this so I did not test the balanced output. What a nightmare these companies have created with each having their own connector for this use.
Starting with output impedance, it is 3.2 ohms:
I would have liked to see around 1 ohm which is the best case situation for my testing. As it is, it not too bad but with balanced it likely doubles to 6 ohm so be careful with such headphones that have uneven frequency response below 60 ohm.
The most important test is power versus distortion+noise. Here is the output for 300 ohm load representing high impedance headphones:
This is unacceptable on multiple front. First, there is just not enough power here for a desktop, mains operated amplifier. Voltage limiting causes this which one could understand if an amp is USB powered but not here. And of course, we have our early distortion rise. If you turn down the levels to avoid this, you are down to 0.1 milliwatts of power!
The situation doesn't improve with 33 ohm load that emphasizes current delivery:
In this price range I like to see power north of 1 watts, not 0.27.
The only good news is channel balance and volume control granularity which is excellent:
You have some 95 dB of attenuation before you lose any channel matching!
Dynamic range at 50 millivolt output is 75 dB:
This puts it in the middle of the pack of the few tested so far:
Listening Tests
As usual I started my listening tests with the Sennheiser HD-650 headphone. Here, power was barely adequate (by desktop DAC standards). Alas, the sound was also harsher at full volume. Dialing down to 92 on the volume control made that go away but then there was not enough volume there.
Switching to much lower impedance Hifiman HE-400i produced more loudness but also, more harshness. There was also lack of detail. To confirm, I switched to my Topping DX3 Pro and the difference was remarkable. The bass, resolution, separation of notes, etc. was in an entirely different class. The Topping was so good I did not want to go back to listening to the Ayre CODEX.
Conclusions
The Ayre CODEX starts with good pedigree and a price that biases one to expect great performance. Sadly we have the opposite. On both objective and subjective front and across wide swath of measurements, the CODEX disappoints. The company says they spent four months conducting listening tests to arrive at these circuit topologies. I would have suggested they spend five minutes in a controlled listening test with a low end product done right to realize they were re-inventing the wheel and make it much less round.
And then we have the high price and ordinary looks of the product here. With high-end products, you usually have great industrial design but not here.
The Ayre CODEX is a great example of a company with excellent engineering talent, misapplying that to chase wrong ideas in audio. It is all an effort to have a differentiated checklist and damn true performance of the product.
When I finish my audio reviews with listening tests, i usually let the unit keep playing while I type up the review. Regardless of performance, the products are usually good enough to let the music fidelity come through unchanged. This is one of the few cases that I did not do that. I was so happy to go back to my Topping DX3 Pro.
------------
As always, questions, comments, recommendations, etc. are welcome.
This is how rich I like to be: a hollywood executive would stay at the same hotel I would when I used to go to Los Angeles. A story was told that he dropped a quarter once and the parking attendant saw it and gave it to him. As a tip, the executive gave him a $5 tip! So please donate money so I can do the same:
Patreon: https://www.patreon.com/audiosciencereview), or
upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
I am puzzled by the look of the CODEX as it doesn't match the rest of the Ayre products:
If you are not familiar with Ayre, it is a US company and is probably best known by its late founder, Charles Hansen. He has been active on forums for years. The company is in high-end audio business so my expectations were high on this DAC and headphone amplifier.
I was pleased to see a set of balanced XLR connectors on the unit and inclusion of mains power supply despite its rather diminutive size:
Then again, closeness of the mains input gives me some pause regarding its noise immunity for the RCA outputs.
In use the CODEX runs very warm. The marketing material talks about push-pull, discrete power supply which likely is the cause (it is a little power amplifier that works on DC).
I did all of my testing with USB as usual. The CODEX was plug-and-play with Windows 10 Creators edition.
Audio Measurements
As usual I start with the same dashboard profile (test settings) for all DACs. Here is the performance with RCA output:
The unit is in "DAC mode" meaning volume control is out of the loop. I was disappointed to see the output voltage not reaching nominal 2 volts. But the biggest disappointment was in the distortion department. 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. I thought something may be broken here so I ran the same test with balanced XLR output:
Mains leakage is reduced but that is not our problem. Our problem are those massive harmonic distortion spike with the 3rd harmonic at 3 kHz reaching way up to -68 dB. As such, by itself sets the SINAD.
I could not believe my eyes. I was ready to take out another DAC to make sure the analyzer was not broken. Then I read some write-ups and figured out the issue is with the design:
Ah, tell me it isn't so. Another believer in voodoo audio and imagined problems with feedback. Electronic parts are not linear. While we try to operate them in the most linear version, they do not have a 1:1 output with respect to input. Feedback was invented almost a century ago to deal with this problem. A small amount of the output is sampled and compared to the input. If they don't agree, the amplifier changes its gain to make them so. This happens extremely fast. There is no time delay as lay people imagine (in audio context). Without feedback I dare say you wouldn't be seeing these characters on your screen! Or make phone calls. Or stream audio/video. It is fundamental to good engineering.
These imagined improvements without feedback are just that: imagination. Anyway, let's continue testing.
The SINAD figure naturally lands the CODEX near the bottom of the pile of DACs out of some 125 tested so far:
Right there with Audio-gd which likewise schews negative feedback.
Jitter performance is good although not as perfect as I like to see in this price range:
Linearity test showed the clear benefit of balanced output and its rejection of mains and its harmonics:
With RCA output (in blue) there is a spike at 180 Hz (third harmonic of 60 Hz in US). My test frequency is 200 Hz and the filtering I use for Linearity test does not exclude that. This means that at low amplitudes (all the way to the left), the mains harmonic has significant amplitude and pollutes the final output. A larger case with better separation from RCA outputs would have helped this possibly. For now, use balanced XLR outputs.
For people who say single tones don't represent music, here is our 32-tone test signal getting subjected to CODEX's non-linearities:
As expected, we see a ton of "grass" in the skirts of our main tones instead of silence. These conspire to hide low level detail in music and cause exaggeration of high frequencies in music, making it more harsh. How anyone can argue this adds more "rhythm" is beyond me.
Intermodulation distortion shows how lack of feedback causes rise in distortion as we ask the DAC to produce higher and higher voltages:
And here is the dynamic range:
Strange to see worse dynamic range with balanced output and such a large drop.
Headphone Amplifier Tests
There are both unbalanced (1/4 inch) headphone jack and "balanced" dual 3.5 mm. I have three other configurations in my balanced test fixture but don't have this so I did not test the balanced output. What a nightmare these companies have created with each having their own connector for this use.
Starting with output impedance, it is 3.2 ohms:
I would have liked to see around 1 ohm which is the best case situation for my testing. As it is, it not too bad but with balanced it likely doubles to 6 ohm so be careful with such headphones that have uneven frequency response below 60 ohm.
The most important test is power versus distortion+noise. Here is the output for 300 ohm load representing high impedance headphones:
This is unacceptable on multiple front. First, there is just not enough power here for a desktop, mains operated amplifier. Voltage limiting causes this which one could understand if an amp is USB powered but not here. And of course, we have our early distortion rise. If you turn down the levels to avoid this, you are down to 0.1 milliwatts of power!
The situation doesn't improve with 33 ohm load that emphasizes current delivery:
In this price range I like to see power north of 1 watts, not 0.27.
The only good news is channel balance and volume control granularity which is excellent:
You have some 95 dB of attenuation before you lose any channel matching!
Dynamic range at 50 millivolt output is 75 dB:
This puts it in the middle of the pack of the few tested so far:
Listening Tests
As usual I started my listening tests with the Sennheiser HD-650 headphone. Here, power was barely adequate (by desktop DAC standards). Alas, the sound was also harsher at full volume. Dialing down to 92 on the volume control made that go away but then there was not enough volume there.
Switching to much lower impedance Hifiman HE-400i produced more loudness but also, more harshness. There was also lack of detail. To confirm, I switched to my Topping DX3 Pro and the difference was remarkable. The bass, resolution, separation of notes, etc. was in an entirely different class. The Topping was so good I did not want to go back to listening to the Ayre CODEX.
Conclusions
The Ayre CODEX starts with good pedigree and a price that biases one to expect great performance. Sadly we have the opposite. On both objective and subjective front and across wide swath of measurements, the CODEX disappoints. The company says they spent four months conducting listening tests to arrive at these circuit topologies. I would have suggested they spend five minutes in a controlled listening test with a low end product done right to realize they were re-inventing the wheel and make it much less round.
And then we have the high price and ordinary looks of the product here. With high-end products, you usually have great industrial design but not here.
The Ayre CODEX is a great example of a company with excellent engineering talent, misapplying that to chase wrong ideas in audio. It is all an effort to have a differentiated checklist and damn true performance of the product.
When I finish my audio reviews with listening tests, i usually let the unit keep playing while I type up the review. Regardless of performance, the products are usually good enough to let the music fidelity come through unchanged. This is one of the few cases that I did not do that. I was so happy to go back to my Topping DX3 Pro.
------------
As always, questions, comments, recommendations, etc. are welcome.
This is how rich I like to be: a hollywood executive would stay at the same hotel I would when I used to go to Los Angeles. A story was told that he dropped a quarter once and the parking attendant saw it and gave it to him. As a tip, the executive gave him a $5 tip! So please donate money so I can do the same:
Patreon: https://www.patreon.com/audiosciencereview), or
upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).