Naim CD 3.5 CD player Measurements
All the measurements on the Naim CD 3.5 were performed using the FlatCap power supply for the player's analog stages. The maximum output level at 1kHz was 2.184V (left channel), which was 0.76dB higher than the CD standard 2V and a negligible 0.04dB higher than the right channel. The player's output was noninverting and sourced from a very low 2.5 ohms in the midrange. This figure dropped even further at high frequencies, to 1.5 ohms at 20kHz, but increased to a moderate 773 ohms below 20Hz, presumably due to a finite-sized coupling capacitor. With a preamp having a low input impedance (less than 2k ohms or so), this will result in a somewhat lean balance, but the only models I can think of like this are the Rowlands. This will not be typical, therefore.
The Naim handled damaged discs superbly well. Track 34 on the Pierre Verany test CD has a gap in the data 2mm long; the Naim negotiated this without dropouts, and even managed to play some of track 35, which has an enormous amount of missing data—some 2.4mm worth!
The CD 3.5's response at 0dBFS (measured with the CBS Test CD) is shown in fig.1 (top traces). The excellent channel matching can be seen, as can some small passband ripples and a slight (-0.6dB) rolloff at the top of the audio band. The bottom pair of traces show the Naim's response with pre-emphasized data. There is a maximum error of -0.7dB in the low treble, which will make that tiny proportion of emphasized discs sound too laid-back. The channel separation (not shown) was better than 100dB below 2kHz, which is superb, with a slight decrease above that frequency, to 85dB or so at 20kHz.
Fig.1 Naim CD 3.5, frequency response at 0dBFS (top) and de-emphasis response (bottom) (right channel dashed, 0.5dB/vertical div.).
Fig.2 shows a spectrum of the player's output while it decoded data representing a dithered 1kHz tone at -90dB. There is a very slight negative level error apparent, as well as a sniff of second harmonic, but the trace is otherwise superbly free from noise and spuriae. Extending the spectrum bandwidth to 200kHz and driving the CD player with "digital black" data gave the trace shown in fig.3. The ultrasonic noise level is low, and while there is a blip in the left channel at 2kHz, this might well be due to the Audio Precision test gear rather than to a DAC idle tone in the Naim.
Fig.2 Naim CD 3.5, spectrum of dithered 1kHz tone at -90.31dBFS, with noise and spuriae (16-bit data, right channel dashed).
Fig.3 Naim CD 3.5, spectrum of digital silence, with noise and spuriae (16-bit data, 1/3-octave analysis, right channel dashed).
Regarding linearity, the Naim's level error (fig.4) remained insignificant to below -110dBFS. Its reproduction of an undithered 1kHz waveform at -90.31dBFS (fig.5) was correspondingly accurate, with clearly visible transitions between the three digital levels describing the waveform.
Fig.4 Naim CD 3.5, left-channel departure from linearity (2dB/vertical div.).
Fig.5 Naim CD 3.5, waveform of undithered 1kHz sinewave at -90.31dBFS (16-bit data).
The CD 3.5's analog stage appears to be bombproof: fig.6 shows the spectrum of the player's output reproducing a full-level 61Hz tone from the CBS Test CD. The only harmonics visible are the second, third, fifth, and thirteenth, and these are all approaching the -100dBFS level! Similarly, the Naim's output spectrum while reproducing an equal mix of 19kHz and 20kHz tones, the combined signal just reaching 0dBFS (fig.7) was superbly free from significant intermodulation products.
Fig.6 Naim CD 3.5, spectrum, DC-1kHz, 61Hz at 0dBFS (linear frequency scale, 20dB/vertical div.).
Fig.7 Naim CD 3.5, HF intermodulation spectrum, DC-22kHz, 19+20kHz at 0dBFS (linear frequency scale, 20dB/vertical div.).
I used the Miller Audio Research analyzer to look at the Naim's jitter performance. This drives the CD player under test with data representing a high-level 11.025kHz tone and a 229Hz tone toggling the LSBs on and off, while it performs a high-resolution FFT analysis—32,768 points, with 64 FFTs averaged—to examine the noise floor in the analog domain.
Like the sample of the CD 3.5 reviewed by Paul Miller for the December '97 issue of the English magazine Hi-Fi News & Record Review (p.42), our sample featured a high clock error: +629ppm. This means that a 20kHz tone will actually be reproduced as a 20,012.6Hz tone; while probably imperceptible, this represents inadequate quality control on the master crystal used by Naim. The actual jitter level was very low, at 207.8 picoseconds (ps). Fig.8 shows the spectrum of the Naim's noise floor for 3.5kHz either side of the 11kHz tone. (The grayed-out spectrum is that of the Meridian 508.24, taken under identical conditions. The Meridian has one of the lowest jitter figures we have measured: 144.2ps.)
Fig.8 Naim CD 3.5, high-resolution jitter spectrum of analog output signal (11.025kHz at -10dBFS with LSB toggled at 229Hz). Center frequency of trace, 11.0125kHz; frequency range, ±3.5kHz. Grayed-out spectrum is that of the Meridian 508.24.
Don't let the scaling of this graph mislead you—the Naim has an extremely low noise floor, even if it is around 6dB higher overall than that of the Meridian. Some data-related jitter sidebands can be seen (red markers 1 and 7), but these are relatively low in level. Perhaps of more subjective importance are the sidebands marked with purple markers 3 and 6: these are jitter components with a fundamental frequency of 650Hz. Paul Miller has conjectured that it is the presence of these jitter components that contributes to the player's "slightly aggressive or forthright balance," in his words.
Interestingly, the Santa Fe sample of the Naim CD 3.5 had slightly higher noise but lower jitter than the sample reviewed by PM. This is shown in fig.9, which shows the jitter spectrum for the Stereophile sample (black) compared with that of the HFN/RR sample (gray). The 650Hz sidebands and their harmonics can be seen to be higher in level in the Miller player, even though the noise floor is almost 3dB lower than Wes's sample.
Fig.9 Naim CD 3.5, high-resolution jitter spectrum of analog output signal (11.025kHz at -10dBFS with LSB toggled at 229Hz). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz. Grayed-out spectrum is that of the HFN/RR sample of the Naim CD 3.5. (Our thanks to Paul Miller for making these data available to Stereophile.)
Summing up: Other than the de-emphasis and clock-frequency errors, these measurements reveal a well-engineered CD player.—John Atkinson