Integrated Amp Reviews

I measured the Masters Series M32 DirectDigital using my Audio Precision SYS2722 system (see the January 2008 "As We See It"). Usually, before I test an amplifier, I precondition it with both channels driving a 1kHz tone at one-third power into 8 ohms for an hour. This thermally stresses an amplifier with a conventional class-AB output stage, but as the NAD is a class-D design, this wasn't appropriate. Even so, before doing any testing I ran it for an hour at a moderate power level, to ensure that it was fully warmed up. Because class-D amplifiers emit relatively high levels of ultrasonic noise that would drive my analyzer's input into slew-rate limiting, all measurements were taken with Audio Precision's auxiliary AUX-0025 passive low-pass filter, which eliminates noise above 200kHz.

Looking first at the analog inputs: With the NAD's volume control set to its recommended maximum of "0.0dB," the maximum voltage gain at 1kHz into 8 ohms measured 26.1dB from the speaker terminals, an input of 140mV producing an output of 1W into 8 ohms. The headphone output offered an insertion loss of 0.2dB: 1kHz at 1V was reproduced as 975mV. Increasing the volume setting to "+10dB" increased the gain by exactly 10dB. The line inputs preserved absolute polarity (ie, were non-inverting), and the input impedance was 10k ohms at low and middle frequencies, dropping slightly to 9.4k ohms at 20kHz.

The headphone output impedance was a very low 1 ohm. With the M32's output optimized for 4 ohm loads, the output impedance at the speaker terminals was 0.17 ohm at 20Hz and 1kHz, but was slightly negative at 20kHz, presumably due to the behavior of the necessary low-pass filter. The action of this filter can be seen in fig.1, where a peak that develops at the top of the audioband increases into higher load impedances. The output is flat into 4 ohms (magenta trace), and starts to droop into 2 ohms (green trace). The peak is highest with our standard simulated loudspeaker (gray trace). Optimizing the M32 for other impedances—as well as 4 ohms, the other settings are 2, 8, and >8 ohms—flattened the top-octave response for the selected load, but increased the height of the peak when the actual load impedance was higher than the nominal setting (fig.2).

I was puzzled by the overall shape of the traces in fig.1, which cut off sharply above 20kHz. But when I looked at the M32's reproduction of a 1kHz squarewave (fig.3), I remembered that the amplifier digitizes its analog inputs and that I'd set the converter's sample rate to its lowest setting, 48kHz. The analog line inputs can also be digitized at 96 or 192kHz. Fig.4 shows the M32's frequency response into 8 ohms with the amplifier optimized for 4 ohms and with all three sample rates; you can see that the output filter's peak reaches its maximum between 30 and 40kHz, with sharp rolloffs just below half of each of the two lower-frequency rates.

Fig.5 shows the effect of the M32's bass and treble controls set to their +10dB and –10dB positions and with the converter sample rate set to 96kHz. The bass control offers a range of ±10dB below 40Hz, the treble control +10/–9dB at 20kHz. Channel separation was good rather than great, at >70dB in both directions below 2kHz but decreasing to 50dB at 20kHz. Without the AP low-pass filter and with no signal present, about 260mV of ultrasonic noise was present at the speaker outputs. With the filter, the analog inputs shorted to ground, and the volume control set to "0.0dB," the wideband, unweighted signal/noise ratio (ref. 2.83V into 8 ohms) measured a disappointing 37dB, due to 41.5mV of residual ultrasonic noise. Restricting the measurement bandwidth to 22kHz increased the ratio to a respectable 73.5dB, and an A-weighting filter increased it further, to 86.6dB. (All ratios are the average of the two channels.) Levels of any residuals at the AC power-line frequency and its harmonics were negligible (fig.6).

The M32 is specified as delivering a maximum continuous output power of >150Wpc into 8 ohms (21.8dBW) or 4 ohms (18.8dBW). At our usual definition of clipping (ie, when the percentage of THD+noise in the amplifier's output reaches 1%), with continuous drive the M32 delivered 190Wpc into 8 ohms (fig.7, 22.8dBW) and 180Wpc into 4 ohms (fig.8, 19.5dBW). Distortion levels at moderate powers were very low, and although the shape of the spuriae waveform at 50Wpc into 8 ohms (fig.9, red trace) is obscured by high-frequency noise, the distortion signature appears to be primarily second-harmonic in nature. This was the case even at high powers into 4 ohms (fig.10). Though higher-order harmonics are visible in this graph, they all lie at or below –100dB (0.001%). Intermodulation distortion at high powers was relatively low, even into 4 ohms (fig.11). However, the noise floor begins to rise above 10kHz, and there is some peculiar spectral content present around 5kHz, though this is extremely low in level.

The phono input preserves absolute polarity and appears to be digitized at a fixed rate of 48kHz. It offers a maximum gain of 61.75dB at the speaker outputs, which is appropriate for moving-magnet cartridges. The input impedance was also appropriate for MM cartridges, at 45k ohms at 29Hz, 39k ohms at 1kHz, but 16k ohms at 20kHz. The RIAA error was both very low and well matched between channels (fig.12), though some odd ripples are apparent in the mid-treble. The phono input's audioband S/N ratio, ref. 1kHz at 5mV, was excellent at 71.5dB, this improving to 83.4dB when A-weighted.

The phono input's overload margins were also excellent, at >16dB across the audioband, and both harmonic distortion and intermodulation distortion were extremely low. However, some odd low-level spectral content was again apparent in the mid-treble region.

A 1kHz digital signal at –12dBFS resulted in an output level of 10.34V into 8 ohms with the volume control set to "0.0dB," which suggests that the M32's gain architecture is well organized. The AES/EBU and coaxial and optical TosLink inputs on the "MDC DD S/PDIF" module all locked to datastreams with all sample rates up to 192kHz. Apple's AudioMIDI app indicated that the USB input accepted 24-bit integer data with sample rates up to 192kHz. Apple's USB prober utility identified the amplifier as the "M32" from "NAD," and indicated that the USB input operated in the optimal isochronous asynchronous mode.

The NAD's impulse response with 44.1kHz data (fig.13) indicates that the reconstruction filter is a conventional linear-phase type, with time-symmetrical ringing to either side of the single sample at 0dBFS. With 44.1kHz-sampled white noise (fig.14, red and magenta traces), the M32's response rolled off sharply above 20kHz, reaching full stop-band suppression by half the sample rate (vertical green line). An aliased image at 25kHz of a full-scale tone at 19.1kHz (blue and cyan traces) can't therefore be seen, though the noise floor starts to rise in the top audio octave and continues to rise at ultrasonic frequencies. The distortion harmonics of the 19.1kHz tone are visible above the ultrasonic noise floor, the second harmonic being the highest in level at –64dB (0.06%).

When I examined the NAD's digital frequency response with S/PDIF data at 44.1, 96, and 192kHz, the results were the same as shown in fig.4 (fig.15). When I increased the bit depth from 16 to 24 with a dithered 1kHz tone at –90dBFS (fig.16), the noise floor dropped by almost 24dB, meaning that the M32 offers 20 bits' worth of resolution, which is excellent. With undithered data representing a tone at exactly –90.31dBFS (fig.17), the three DC voltage levels described by the data were well resolved and the waveform was perfectly symmetrical. With undithered 24-bit data, the result was a clean sinewave (fig.18).

With full-scale data representing equal-level tones at 19kHz and 20kHz (fig.19), intermodulation products are relatively low in level but the rise in the noise floor in the top octave is visible. The NAD's rejection of word-clock jitter with 16-bit AES/EBU data (fig.20) was superb, with all the odd-order harmonics of the LSB-level, low-frequency squarewave at the correct levels (fig.19, sloping green line). With 24-bit J-Test data (fig.21), no jitter-related sidebands were present above –130dB (0.00003%) via the AES/EBU, coaxial S/PDIF, and USB inputs.

NAD's M32 packs a lot of well-engineered performance into its relatively small, discreet case.—John Atkinson