Laboratory test report
The measured frequency response of the Yamaha CD-S300 was extraordinarily flat, as you can see from the tenth graph accompanying this report. The first thing to note when looking at Graph 10 is that each horizontal line represents a difference in level of just 0.2dB, meaning that the total range of the graph, from top to bottom, is only one single decibel (1dB). So, although the Yamaha CD-S300’s frequency response appears to get a bit ‘squiggly’ and roll off above 10kHz, that roll-off is from the 0dB reference line down to just –0.14dB at 20kHz. This puts the Yamaha CD-S300’s measured ‘normalised’ frequency response at 20Hz to 20kHz ±0.08dB. Note also that the player’s low-frequency response extends down to 2Hz… this just isn’t shown on the graph, whose lowest display frequency is 20.56Hz.
Another view of the Yamaha CD-S300’s frequency response can be seen in Graph 7, which shows the player’s response to an input signal at 630 impulses per second. On this you can see the low-frequency extension, but also the frequency response above 20kHz, which rolls off to be more than 80dB down at 25kHz… not an issue as the highest frequency that (theoretically) can be recorded on a CD is 22.05kHz (i.e., half the sampling frequency of 44.1kHz).
Channel separation was excellent, with Newport Test Labs reporting measured results of 124dB at 1kHz, 117dB at 16Hz, and 118dB at 20kHz. Apart from being excellent results, these results are far better than would ever be required to give maximal channel separation and ideal stereo imaging. Channel phase errors were also very low, with a ‘worst-case’ result of just 0.11° at 20kHz. Group delay was typical for an ‘off-the-shelf’ delta-sigma DAC.
You can see from the tabulated figures that Yamaha has gone to the trouble of implementing the de-emphasis circuit that’s fitted to the DAC (but often not used by manufacturers, due to the added cost of the support circuitry), which means that the CD-S300 will correctly add de-emphasis to pre-emphasized CDs. There can’t be too many of these CDs around, as manufacturers stopped this practice in around 1989, but if you have super-old CDs in your library, you can be assured that they’ll play back correctly… particularly since the de-emphasis circuitry that is fitted to the Yamaha CD-S300 is so precise, not least at 20kHz, where Newport Test Labs measured an error of a minuscule 0.009dB.
Linearity error was a little below average, particularly at very low recorded levels, but these errors would not be audible when listening to music.
Distortion at 0dB is shown in Graph 1 and you can see that there’s a second harmonic at –100dB (0.001% THD) and a third at –95dB (0.0017%). Other harmonically-related distortion components are also visible, but they’re all down around –120dB (0.0001%) and would be inaudible. The overall noise floor is very low, as you can see, mostly more than 130dB down, but the noise floor rises at very low frequencies (extreme left of graph), which is probably due to some mains-frequency hum (50Hz) and components, and more than likely responsible for the only average tabulated overall signal-to-noise ratios measured by Newport Test Labs of 99dB unweighted and 102dB A-weighted.
Distortion at –20dB recorded level is shown in Graph 3 and the distortion is not only a little higher than I might have expected, but also exclusively comprised of odd-order harmonics and converter artefacts. However all the components are sufficiently low in level that they would not be audible, even if you play your music at extremely high volume levels. The highest of the components visible on the graph is the third harmonic, and this is positioned at a level of –108dB, or just 0.0003% THD.
At a recorded level of –60dB, distortion increases again, related to recorded level, and is again exclusively odd-order. This time granulation noise becomes quite prominent, as it usually does at such low levels. Again, however, the highest (third harmonic) distortion component is at –101dB (0.0008%), the other components are down around –110dB (0.0003%) and the granulation noise is down at around –130dB.
Down at a recorded level of –90/91dB (Graphs 5 and 6), the undithered signal (Graph 5) shows typical converter errors, but even when the signal is dithered, there’s a significant second harmonic component remaining in the output, which would suggest an older DAC architecture.
Intermodulation distortion was low for both tested types, with CCIF IMD shown in Graph 8. You can see the unwanted regenerated difference signal at 1kHz is more than 100dB down (0.001%), as are the two sidebands at 18kHz and 21kHz. These are good results.
The oscillograms showing the Yamaha CD-S300’s response to a 1kHz square wave test signal and an impulse exhibit the typical
time-reversed ringing of a standard delta-sigma oversampling filter.
Newport Test Labs measured the output voltage of the Yamaha CD-S300 when playing back a 1kHz signal recorded at 0dB as 2.11005-volts from the left channel line output and 2.1336-volts from the right channel, so essentially only a 0.01V difference, equating to a channel balance of 0.09dB, which is excellent.
Newport Test Labs was not able to measure power consumption in standby mode, because the CD-S300 does not have a standby mode. If, as a result, you leave it switched on (but not playing) when you’re not using it, it will draw a continuous (though admittedly rather insignificant) 7.47-watts from your mains power supply. When you’re actually playing a CD, power consumption increases to just over 10-watts.
Other than the fact that it didn’t appear to have a stand-by circuit, the Yamaha CDS300 delivered solid performance on the test bench. Steve Holding