Laboratory test report

The output voltage from the SA-14SE was plenty high enough to drive any integrated amplifier, even high enough to drive a power amplifier directly if that’s your preferred setup. As you can see from the tabulated results, Newport Test Labs measured 2.37 volts on the left channel, and 2.35 volts on the right, with the very slight difference meaning that there was a channel imbalance of 0.06dB. Not enough to be audible.

Channel separation was impressively high, even up at 20kHz, where CD players typically find it tough to deliver, at which frequency the SA-14SE’s result was in triple digits (101dB). As you can see from the tabulated figures, separation improved to 125dB at 1kHz and 126dB at 20Hz. All these figures are light-years more than will ever be required to deliver maximal stereo performance.

Channel phase errors were also low 0.53° at 20kHz, 0.02° at 1kHz and non-existent down at 20Hz (0.00°).

Linearity errors were very low, once more, as you can see from the tabulated results, varying from non-existent (at –60dB) to a maximum of 0.09dB (at –80.70dB).

Overall THD+N at 0dB was 0.05% and you can see the output spectrum at the level (Graph 1) shows second and third harmonic distortion components at –108dB (0.00039%) and –105dB (0.00056%), plus a fifth at –115dB (0.00017%). Some other components are present, but they’re all at or more than 125dB (0.00005%) down. These all appear to be analogue components caused by driving the output stage to it maximum level, which is never going to be the case when playing music, which is recorded at much lower levels to ensure there is no clipping on peaks. This would seem to be evidenced by Graph 2, which shows output at –10dB recorded level, which is still quite high. On this graph, you can see there’s only a single harmonic distortion component (the second harmonic) at –114dB (0.00019%). Reducing the recorded signal level a further 10dB, to –20dB (Graph 3) sees even this single component drop further in level, to –126dB (0.00005%).

At –60dB (Graph 4) we begin to see artefacts from the digital-to-analogue converter in the noise floor, but all of these are more than 120dB down, or less than 0.0001%.

Distortion performance at extremely low recorded levels was exceptionally low, as you can see from Graph 5, which shows an undithered sine wave at a level of –91.24dB. There is a second harmonic at –108dB (0.00039%), followed by the next three higher-ordered harmonics at –118dB (0.00012%). All others are below –120dB (0.0001%). Although these are all low, they only occur at all because the test signal is not dithered, whereas music is dithered. The result with a dithered test signal at almost the same level is shown in Graph 6 and you can see there is no distortion visible in the output of the SA-14S1 at all—just the noise floor, and that’s at –140dB right across the audio band. This is a superbly low noise floor that’s 10dB to 20dB lower than I normally see. Total overall wideband noise was 107dB unweighted, improving to 115dB A-weighted—both excellent results.

CCIF IMD was measured for both settings of the Marantz’s two user-adjustable low-pass filters. The result with Filter 1 is shown in

Marantz’s SA14S1 SE returned outstanding performance in all three of its operating modes...

Graph 7 and you can see high-level IMD components at 24kHz and 25kHz, a regenerated 1kHz signal plus signals at 14kHz and 16kHz. With Filter 2 selected (Graph 8), both the high-level IMD components and the in-band signals disappeared completely. Only the regenerated 1kHz signal remained, and this was down at –118dB (0.00012%).

The same general level of performance was demonstrated with a single 20kHz sine wave at maximum recorded level (0dB), as you can see in Graphs 9 and 10. When using Filter 1 there is a high-level signal at 24kHz as well as in-band components, whereas when using Filter 2 these disappear completely.

You can see the effect of the two filters on the Marantz SA-14S1 SE’s frequency response in Graph 11, with Filter 1 rolling the response off quite slowly, whereas Filter 2 cuts it off incredibly steeply. Neither filter’s action starts cutting in until above 14kHz. The Marantz’s frequency response is ruler flat from 2Hz up to 2kHz and only –0.1dB down at 14kHz. Overall, with either filter, the Marantz’s frequency response is 2Hz to 20kHz ±0.1dB.

The two different filter slopes are shown rather more graphically in Graphs 12 and 13, where you can clearly see the shallow slope of Filter 1 (Graph 12) vs the sharp cut-off of Filter 2 (Graph 13).

Newport Test Labs also tested the Marantz SA-14S1 SE’s performance as a DAC, using AES-17 standard digital test signals (24-bit/48kHz). The tabulated results reveal excellent performance across all the standard tests, with non-linear interchannel crosstalk figures in particular being commendably low. Channel separation was, rather curiously, just a little bit less than when using 16-bit/44.1kHz test signals, as you can see in Graph 14, with a little bit of variance depending on whether the measurement was L–R or R–L, but either way, separation was greater than 100dB below 6kHz and better than 90dB at 20kHz. Interchannel phase, on the other hand (Graph 15) was virtually identical to the results measured by Newport Test Labs using 16/44.1 test signals.