LABORATORY TEST REPORT

Newport Test Labs measured the output of the Arcam CDS50 as being a bit more than 2.23 volts from the unbalanced outputs, and a bit over 4.45 volts from the balanced outputs, so it’s a true balanced output. Channel balance was an outstanding 0.0001dB, which is exceptional, while channel separation was also outstanding, with a best result of 148dB at 1kHz, and a ‘worst’ of 133dB at 20kHz. I’m not sure that I’ve seen any CD or SACD player (or DAC) have such good channel separation at 20kHz. It shows excellent circuit engineering.

The frequency response of the Arcam CDS50 was measured for the CD section using a 16-bit/44.1kHz test signals and for the DAC using 48kHz/24-bit test signals. The responses were exceptionally flat, as you can see from the graphed results. For the CD section, the response was 20Hz to 20kHz ±0.13dB. Inter-channel phase was also exceedingly low. The phase error results for the DAC section have been graphed, so you can see for yourself. For the CD section, there was a 0.02° error at 16Hz, no error at all at 1kHz, and a 0.05° error at 20kHz. These errors would be totally inaudible, even under ideal circumstances.

Graph 1 shows THD at 1kHz for a 0dB recorded signal from CD, and although there are some distortion components visible, they’re all more than 100dB down (so less than 0.001%) and this is a ‘worst case’ signal, since no commercially recorded CDs will have signals recorded on them at such a high (0dB) level. In fact, most of these distortion components were caused by the Arcam’s own output stage being pushed to the limit by the 0dB signal, because you can see in Graph 2, where the level of the test signal has been reduced to –10dB that absolutely all the distortion components have disappeared except for the third harmonic, which is a converter-related component, and this harmonic is sitting down at –117dB (0. 0.00014% THD).

At a recorded level of –60dB, the output of the Arcam showed the usual granulation noise one would expect when a DAC was converting a low-level undithered signal. Granulation noise is also present in Graph 4, where the test signal is an even-lower –80.59dB (again not dithered) and you can see that in addition to the granulation noise, the DAC has introduced some odd-order harmonic distortion. So what happens when the test signal is dithered? You can see this in Graph 5, which shows the Arcam CDS50 reproducing a dithered 1kHz test signal recorded at –80.70dB. There’s just the test signal at the left of the graph, then no distortion whatsoever, just a clean noise floor down at –140dB. Excellent performance.

The same effect is demonstrated in Graphs 6 and 7 which show an undithered signal at –89.46dB (Graph 6) and a dithered signal at –90.31dB (Graph 7). You can see the distortion that’s visible in Graph 6 disappears completely in Graph 7. However, you can also see that when the test signal is not dithered, the noise floor is lower than it is when the signal is dithered. This is the ‘penalty’ you pay for dithering a digital signal—an increase in the level of the background noise—but when that background noise is 140dB down, as is the case here with the Arcam CDS50, the increase is not important, because even at the higher level it will be totally inaudible. Remember that it’s only test CDs and digital test signals that are not dithered, precisely so they can show DAC performance—all digital music signals will be dithered. You can read an excellent article about what dither is and how it works at www.avhub.com.au/dither

Graph 8 shows the output of the Arcam CDS50 when it’s reproducing a 20kHz signal recorded at 0dB. You can see there’s a little interaction with the sampling frequency that results in unwanted signals up around 24kHz, 36kHz and 40kHz, but all the signals visible on the graph are close to or more than 100dB down and at frequencies that are in any case so high they’re inaudible (and most especially at these low levels!). The same was true when the Arcam CDS50 was reproducing a CCIF IMD test signal (Graph 9), however due to the lower recorded level, the sidebands and sampling artefacts are even lower in level, more than 110dB down (0. 0.00031% THD). The unwanted signal that’s regenerated down at 1kHz (being the difference signal between the 19kHz and 20kHz test signals) is sitting 118dB down (0. 0.00012 % THD).

Arcam is using a very steep high-cut filter on the output, which is shown in Graph 10, which uses a 630 pulse per second test signal to look at the filter slope.

You see there’s slight roll-off from 20kHz to 22.5kHz, then a ver y steep roll-off thereafter. The in-band frequency response of the CD section of the Arcam SD50 is shown in Graph 11.

You can see that it’s ruler flat out to 1kHz, then there’s a tiny roll-off to be 0.09dB down at 10kHz, then a further rolloff to be 0.26dB down at 20kHz.

Unlike many modern players, the Arcam CDS50 has a de-emphasis circuit fitted, so it will be able to accurately replay very old CDs (by ‘very’ old, I mean CDs recorded prior to 1990). The circuit is very accurate too, with only a 0.007dB error at 16Hz, and only a 0.008dB error at 1kHz. Even at 16kHz, the circuit’s error amounts to just 0.195dB, which is such a tiny error that it would be completely inaudible.

Newport Test Labs measured the signalto-noise ratio of the Arcam CDS50 as 113dB A-weighted, which is excellent.

Linearity error was vanishingly small, as you can see from the figures tabulated in the ‘Test Result’ table, with zero error at –80.59dB, only 0.01dB at –60dB and between 0.02dB and 0.05dB at all other levels tested except –80.70dB, where it was 0.09dB.

If you’re also using the CDS50 as a DAC, you’ll find it delivers excellent performance. The inter-channel phase response (Graph 12) is almost perfect, barely deviating from the 0dB graph line from 10Hz right up to 20kHz. Interestingly, channel separation didn’t seem to be quite as good as was demonstrated in the CD test, but it’s still more than 100dB right across the audio band, so it’s far better than will ever be necessary to ensure perfect channel separation and properly focused stereo imaging.

The signal-to-noise ratio of the digital input, on the other hand, returned a better result than that measured for CD, with a figure of 121.531dB CCIR-RMS weighted.

No matter whether you’re using it as a CD player, as an SACD player or as a DAC, the Arcam CDS50 will deliver state-of-the-art performance. Steve Holding