For someone who has loved the compact disc for more than forty years, it’s kind of puzzling that one of the most prominent “features” listed for a CD player is “gapless playback”. For the first twenty years of the format, I never came across a CD player that offered anything but gapless playback. My first experience to the contrary was a (relatively) low-cost Chinese DVD player. I was stunned that my CDs with run-on tracks would have a break between said tracks.
So I guess it’s saying something that the description of the Cambridge Audio AXC35 CD Player on the company’s website opens, “AXC35 is an elegant, gapless playback CD player featuring …” I can confirm that it does indeed have that “feature”. And as we’ll see, any high-quality audio system lacking a CD player would be significantly enhanced by including the Cambridge Audio AX35.
First, some basics:
Information
- Frequency response: 20 to 20,000 hertz +/-0.4dB
- Dimensions: 430mm w x 305mm d x 75mm h
- Weight: 4.3 kilograms
- Accessories: Remote control, Quick start guide, Power cable, Batteries for remote
- Manufacturer Website here.
- User manual here.
- Price: $689
- Available here.
About the Cambridge Audio AXC35
I feel that the Cambridge Audio AXC35 looks the way that a CD player ought to look. It has a not-quite-silver metal fascia, a CD tray that opens outwards with a display underneath and a row of control buttons along with an IR receiver. That’s it on the front.
At the back are a pair of RCA outputs for analogue audio and an RCA coaxial output for digital audio, along with a figure-eight power socket for the included cable. As is Cambridge Audio’s practice, the labels on the rear connections are printed conventionally underneath them, and also upside down above them. That makes it easy to identify them when you’re leaning over the top with the equipment is in place.
It these days of lightweight gear, the AXC35 feels substantial at 4.3 kilograms.
Inside it delegates digital to analogue conversion to a Wolfson Microelectronics WM8524 chip. This is a fairly old-school processor – its spec sheet has a 2011 copyright. It supports 24 bits and signals to 192kHz sampling and offers a 106dB A-weighted signal to noise ratio. Does that seem modest compared to today’s state-of-the-art DAC chips, some of which offer better than 120dB SNR? I suppose, but this is a CD player. It is limited to 16-bit digital audio, and the theoretical best SNR on offer from 16 bits is 96dB.
It has the output circuitry on-chip, providing 2 volts RMS full scale. I don’t know if this unit uses that or has its own analogue coupling circuitry.
Even though it is fairly basic in terms of functionality – no CD text, for example – the unit does all that you’d expect of a CD player, plus a little extra. It will play CD-Rs with MP3 (I checked this) and WMA (I didn’t check this) files. It will not play losslessly compressed audio, such as FLAC.
The remote control also works with the company’s AXA35 integrated amplifier but nonetheless has a good range of keys devoted to the CD player, including numeric keys for jumping directly to selected tracks, repeat and random playback options, and even playback programming, with the tracks you select.
I suffered a little remote envy, because the remote with my Cambridge Audio CXC Series 2 CD transport only has the most basic controls. Unfortunately, the AXC35 remote uses different codes so there was no point buying one for my unit.
Why CDs?
No, the compact disc isn’t dead. Here is why you should have a CD player in your system.
First, you’ll own the music forever. If a Neil Young throws another snit about a podcast of which he disapproves, you’ll still have Harvest to listen to, even as it disappears off Spotify.
Second, CDs are cheap. Over the past five years or so I’ve acquired a very extensive classical collection of music at a cost of between fifty cents and two dollars per CD, courtesy of Vinnies and Salvos. At that price, you can give them back the ones you don’t like.
Third, I challenge you to listen to Wendy Carlos’ Switched-On Bach. This 1968 album established the synthesiser as a legitimate musical instrument, not a mere effects machine. It was the first classical recording in a decade to make it to the Billboard 200, and it became platinum. To check it out, you’re going to have to buy it second-hand on vinyl or CD because almost none of Wendy Carlos’s oeuvre is available on streaming, including her soundtrack to A Clockwork Orange.
Using and listening to the Cambridge Audio AXC35 CD player
I used the CD player extensively in two ways. One was with its analogue output plugged into an RME ADI-2 Pro FS R Black Edition ADC/DAC. This has a highly competent headphone amplifier, and in this mode I mostly used Final Audio D8000 headphones, but also Focal Elear and FiiO FT7 headphones.
And I also used it with my main system, with an Advance Paris preamplifier, a now-superseded Simaudio Moon 330A power amplifier and DynAudio Contour 20i loudspeakers.
In summary, the music produced by the Cambridge Audio AXC35 with any of these options proved to be utterly awful, or wonderfully sublime, or somewhere in between. It all depended upon the CD. The AXC35 was perfectly faithful to what was on the CD.
For example, as my fingers are typing I have the FiiO headphones on my head and Chandos CD 9288 spinning in the AXC35. This is a recording of, amongst other things, the Rodion Shchedrin reworking of Bizet’s Carmen into a percussion-intensive piece. I first heard this version of the piece (my grandmother gave me a vinyl copy of a behind-the-iron-curtain recording when I was twelve) in a demonstration of a $250,000 loudspeaker system. I bought the CD (new, full price) almost immediately.
This is how a recording should be made. But you can’t listen to it on Spotify, as least as far as my searching would reveal. You need the CD.
Really, you do.
Then I listened to U2’s All That You Can’t Leave Behind, and my previous impressions of a nasty, crunchy mastering were confirmed. Is U2 one of the originators of the Lo-Fi genre? It sounds a bit better with the Final Audio headphones than the loudspeakers or the other headphones, because these hold back the crunchy upper midrange, lower treble. But even with those headphones the sound is kind of thin.
And no, this isn’t a classical vs rock issue. I’ve switched over the Primus’ tales from the punchbowl, and this is almost as clean as the Chandos CD.
What I really want in a CD player is the CD, transparently. And while it may have been around for a few years, that is what the Cambridge Audio AXC35 continues to deliver.
I also played plenty of potentially problematic CDs. One CD won’t play a track on several FiiO portable CD players. No problems here. A couple of years ago I went to Salvos with a mission: buy a bunch of CDs with significant surface marking. These all played flawlessly.
As did the transparent version of Lorde’s most recent CD, Virgin, which the internet claims causes problems with some players (personally, I think that’s from people placing it upside down.) And the “DualDisc” version of Streisand’s Guilty Pleasures. That was a short-lived format with a kind-of CD on one side and a DVD on the other. The problem with the CD side is that the data was more shallowly-placed that the official CD specification provided. But still no problems here.
I think if you come across a disc that the AXC35 won’t play, it probably won’t play in any other player either.
Some Measurements of the Cambridge Audio AXC35
I never expect any obvious measured deficiencies in equipment from quality brands, such as Cambridge Audio, but sometimes there are differences around the edges. And with CD players, I do like to attempt to work out what output filtering strategy that has been adopted.
So let’s start with the frequency response:
What you see there is exactly what I like to see: a flat line extending from 20 hertz to 20,000 hertz. In this case, it fits within a 0.1dB envelope. The cutoff just on 20kHz is very sharp – which is also something I like to see. That suggests the employment of a “Sharp” or “Fast” filter, which generally provides the most accurate reconstruction of the input signal. The transient response was:
This is characteristic of a linear phase (as opposed to minimum phase) filter. Don’t worry about those ripples. They are simply a mathematical artefact of any filter. The test signal I use is designed to make them obvious. No music will ever produce such results. See my article here for the full explanation.
The noise floor is low in the no-signal state. I used 24-bit measurements in this case to see what was happening below the 16-bit noise floor:
The noise is mostly below -132dB, with a small, utterly inaudible, bump at the 50 hertz mains frequency, one at 150 hertz, and a few random ones here and there. The noise totalled up came in at -108.2dB, A-weighted. Clearly, this is the analogue noise floor when there’s effectively no digital signal.
I find the harmonic distortion graph more useful to see what’s happening with noise:
Now we have an actual 16-bit signal and as you can see, the noise floor is very flat, consistently between -123dB and -126dB. That means no surprises, and about what you’d expect from a well-performing CD player. The harmonic distortion levels – they’re the spikes to the right of the 1kHz test signal – are about the usual levels and all below -96dB. Which means, utterly inaudible. The total harmonic distortion of this CD player came in at 0.0033%.
Finally, stereo crosstalk was at -105.1dB.
All these figures are better than claimed in Cambridge Audio’s specifications for the AXC35.
Using the Cambridge Audio AXC35 as a CD Transport
In essence, a CD Transport is simply a CD player without an analogue output section. Instead, it provides the CD’s content in digital format, typically via optical or coaxial digital audio, and sometimes via AES (this uses XLR cables). Many CD players – which of course have analogue outputs – now also provide digital outputs. The idea is that you can use an external DAC to provide even higher performance.
That’s assuming that the digital signal delivered by the CD player is perfect. Unfortunately, that produced by the AXC35 isn’t, quite. I actually wasn’t going to bother to check this. Cambridge Audio has been in the business a long time and, in any case, I’ve previously tested my own Cambridge Audio CXC Series 2 CD transport, and it does indeed deliver a perfect bitstream. The only reason I did check was because I had the gear out and set up to check a different CD player about which I had questions, so I thought why not.
What the AXC35 seems to be doing is adding 1 bit of dither. That is, a signal just one bit high, consisting of quasi-random 1s and 0s, is added to the output. Here’s what it looks like:
Dither is a useful thing to add at times to digital audio in order to reduce certain types of noise and distortion, as I explained a while back here. But it should not be added willy-nilly. If one bit of dither is added at one point in the production chain, and another at another point, then you’re getting 2 bits of dither. Each bit increases the noise floor slightly. Few people, if any, would detect the presence of 1-bit. But we know for certain that many CDs include the addition of dither. Sony’s “Superbit” CDs employ shaped dither to expand the dynamic range.
Obviously, I was very surprised by this, so to make sure I wasn’t messing things up with my measurements, without switching off my computer or audio interface, I used the same cable to plug into the CXC S2, and played the same track on the same CD, and recorded it as a new track in the same Project in Reaper. That one was a perfect match for the CD rip, from which the AXC35 differed by that random bit.
Look, if you plug it into an external DAC, you’re going to get sound that’s very nearly identical to that on the CD, but if you’re buying a CD player with the intention of using it as a CD Transport, I’d suggest looking elsewhere.
Of course, who knows how many other CD players muck around with the digital output? I can only confidently recommend the Cambridge Audio CXC Series 2, because I tested it and it’s perfect.
Intersample Overs?
A manufacturer of professional DACs, Benchmark Audio, makes a big thing of so-called Intersample Overs. Obviously it has products that do not suffer from this. The idea is that some digital audio signals, such as on a CD, are recorded at such a high level that the analogue signal they encode would actually reach above the hard digital limit of 0dBFS. That wouldn’t be a problem if that all there were to it. It would be easily handled so long as the analogue stage of the DAC had about 3dB of headroom above that limit.
The problem arises in the digital domain. Most DACs “oversample”. Sticking with CDs, that means they might convert the 44.1kHz to, say, 176.4kHz (4x oversampling), calculating new digital samples between the source ones. This has certain advantages, in particular when it comes to designing effective “reconstruction” filters, which can be done in the digital domain.
But what if two adjacent samples are so high that the analogue signal between them would have reached higher than 0dBFS? The digital system, if it’s operating with the usual 16- or 24-bit integers, can’t create samples that faithfully follow the curve, introducing distortion.
The ways around this include digitally reducing the signal level by 3dB before upsampling – which raises the noise floor by 3dB – or converting to either 24-bit integer or 32-bit floating before the 3dB reduction. Those have no effect upon the noise floor.
The Cambridge Audio AXC35 does none of these. It simply chops off the top of the signal. Here’s the worst case, using an 11.025kHz test signal (one quarter of the sampling frequency) encoded such that every sample is at 0dBFS:
That creates quite a few high-level harmonic distortion elements, thus:
Which does look pretty alarming. The third harmonic is only about 7dB below the sine wave fundamental, and the other odd-order harmonics are also very high.
So the question is: is this a problem? Benchmark Audio says yes. But they show spectrums that have lots of intermodulation products below the 11.025kHz test signal. And they say that these are audible as a kind of white noise.
I remain uncertain, but I am inclined to say, “No, it’s not a real problem”. First, a maxxed-out 11.025 signal is required to generate that 3dB over which is being chopped off, and that is a ridiculous signal. Nothing approaching that ever appears in real music. Even while Benchmark identifies thousands of intersample overs in some tracks, they point out that most are less than 1dB. 3dB is the theoretical highest. Here’s the AXC35’s spectrum at -2dB (ie. 1dB of overshooting):
That brings the worst distortion products down by 17dB.
Second, that 3rd harmonic is slightly above 30kHz, well outside anything anyone can hear. By its nature, oversamples are more likely with high frequencies than with lower ones. As far as I can work out, the Benchmark claims relate to spurious (non-harmonic) distortion that their measurements reveal, but which I haven’t seen in any of the DACs and devices I have so far measured. Perhaps if fed into a poorly designed amplifier with terrible intermodulation distortion performance, that would be different.
All that said, just in the interests of a full commitment to a pure signal, it would be nice if Cambridge Audio could address this.
Old CDs
One minor weakness was rather surprising to me: the AXC35 didn’t act on the pre-emphasis flag in the CD stream. This is part of the CD specification and is a legacy from the early days of CD when some were encoded with boosted treble. The idea was that a flag in the digital stream told the player to proportionately reduce the treble of these CDs, and thereby reduce any treble noise created due to the analogue to digital and then digital to analogue conversion.
This was a rather pointless idea right from the outset. Nonetheless, a small number of CDs in the first few years of the format were so encoded. There’s no reliable master list of which ones they are. I’ve identified something like three in my collection of several hundred classical CDs, and none in my collection of modern music.
Still, it is odd, especially as the Cambridge Audio CXC CD transport does notice the flag, and feeds it through to the connected DAC, some of which actually act on it! The AXC35 does not pass the flag through to the DAC.
Conclusion
What do they say? “When you’re on a good thing, stick to it.” The Cambridge Audio AXC35 CD player was introduced way back in 2019 and they’re still making them. I expect that’s because buyers are very satisfied with what they have to offer, as indeed was I.
Yes, I’d like that coax to lose the dither, and if possible the inter-sample overs to be cleaned. But in my opinion, neither of these are audible issues. Honestly, I doubt that if you were to spend ten times the cost of the AXC35 on a CD player, you’d get a noticeable improvement in sound. Which is presumably why Cambridge Audio is sticking to it.
