Solid State Power Amp Reviews

Starting with the NAP 62 preamp, I measured a phono gain of 55.7dB, just enough for higher-output "low-output" moving-coils, but more suited to the "high-output" variety. The low-to-moderate phono gain figure suggests the lowest-output moving-coils should be avoided. RIAA error (fig.1) was negligible, but not perfectly flat: a very slight dip (about 0.1dB) between 100–500Hz is evident, along with a 0.2dB rolloff at 20kHz and a steep cutoff below 30Hz. The sharp attenuation slope below 30Hz is fairly typical of European amplifiers that follow the IEC-recommended high-pass rolloff. Interestingly, the rolloff begins at a lower frequency than I have previously measured in other British amplifiers (the Audiolab 8000A, for example), and is very different from the virtually flat extension to 10Hz measured in the Naim NAIT 2 integrated amplifier. Infrasonic problems related to the phono front end should thus be ameliorated by the phono stage's lack of LF extension.

Phono-channel separation was excellent, measuring 95dB at 1kHz, but decreasing to 74dB at 20kHz (L–R). R–L was similar, measuring 88dB at 1kHz, again decreasing with frequency to 72dB at 20kHz (measured with the input shorted). This greater crosstalk as the frequency increases is indicative of capacitive coupling between channels.

Phono-input overload was 550µV at 20Hz, 5.63mV at 1kHz, and 57.5mV at 20kHz, suggesting plenty of headroom with most moderate-output moving-coil cartridges. However, the 62's phono section may be overloaded by very-high-output moving-coils (those in the 2–2.5mV output range) when playing discs with high peak levels. Phono signal/noise ratio was very good at 79.3dB (left) and 78.9dB (right). (All phono measurements were made with the "K"-version boards.)

Moving to the line stage, frequency response (fig.2) was flat, but with a slight 0.2dB rolloff at 20kHz and 0.5dB attenuation at 20Hz. This is contrasted with the Naim NAIT 2 integrated amplifier, whose line stage exhibited a full 2dB of attenuation at 20kHz. Maximum output level was a healthy 8.9V when driving 100k ohms, far in excess of the NAP 140's input sensitivity. Line-stage gain was a moderate 8.2x or 18.3dB. Separation between line-stage channels (fig.3) was 100dB at 20Hz, dropping to 52dB at 20kHz (L–R), measured with the input shorted. Again, the separation decreasing at 6dB/octave as the frequency increases indicates capacitive coupling between channels. THD and Noise was quite low at about 0.02% throughout the audio band (measured at 2V output). Volume-control tracking was excellent, maintaining better than 0.04dB channel balance at any output level between 50mV and 1V.

The NAC 62's output impedance was quite frequency-dependent, measuring a very low 6.3 ohms at 1kHz and 4 ohms at 20kHz, but increasing substantially in the bass to 4400 ohms at 20Hz. A negligible level of DC (3mV) was present at the 62's outputs. Finally, driving the NAC 62 with a positive-going pulse revealed that the preamp is not polarity-inverting.

Looking next at the power amplifier, the NAP 140 clipped (1% THD) at 51.4W/50.1W into 8 ohms (17.1dBW/17dBW), 77W/72.9W into 4 ohms (15.9dBW/15.6dBW), and 97W into 2 ohms (13.9dBW/13.6dBW), with one channel driven at a time by a 1kHz sinewave, shown in fig.4. These figures were substantially reduced, however, with both channels driven: I then measured clipping in the left channel (the slightly more powerful of the two) at 42.4W into 8 ohms (16.3dBW), 61.3W into 4 ohms (14.9dBW), and 70.3W into 2 ohms (12.5dBW). These figures indicate a less than robust power supply. Distortion as a function of frequency at 1W into 8 ohms, 2W into 4 ohms, and 8W into 2 ohms is shown in fig.5.

Next, I measured the 140's power bandwidth into 8, 4, and 2 ohms with both channels driven. This test reveals the amplifier's ability to deliver power at a specified distortion (usually clipping or 1% THD) across the audio spectrum. The Audio Precision System One monitors the distortion level in the amplifier's output and varies the input to the amplifier so that a specified distortion level is reached. It then measures and plots the power output, repeating the process at discrete frequency steps. Using this method, the 140's power bandwidth curves showed a significant drop in output power above 2kHz, especially into low impedances. I found this odd: repeating the test on a Forté Model 4 produced a "normal" curve, indicating the test method was not at fault.

After a discussion with JA about this anomaly, we speculated that because the amplifier is driven nearly to clipping continuously throughout this test, the 140's power supply ran out of reserve after the first few freqency steps, meaning that the voltage rails drooped, lowering the 140's measured maximum output power. Consequently, I measured the output power at discrete frequencies by hand (rather than the System One's automated test), turning off the input signal between frequency steps. (The NAP 140 was also connected to the dummy load with Naim speaker cable for these measurements and driven by the NAC 62 preamplifier.) JA plugged the measured power output figures for 1% THD into the data file, producing the curves of fig.6 (8 ohms bottom, 4 ohms top). These curves correlate very closely with Naim's own measurements and are perhaps more significant than the automated test results considering the transient nature of music.

The power output measurements suggest the NAP 140 would be more comfortable driving moderate rather than low impedances. Unfortunately, the Hales System Two Signature presents a difficult load for the 140: its impedance drops to near 3 ohms for much of the band below 300Hz, and stays around 4–6 ohms through the rest of the spectrum. The power-bandwidth measurements, along with the 140's less than impressive ability to substantially increase its power output into low impedances with both channels driven (delivering 16.3dBW into 8 ohms, 12.5dBW into 2 ohms), suggest that loudspeakers with lowish overall impedances, or those with severe impedance dips, should be avoided. However, into a solid 8 ohms, and at moderate playback levels, the 140 can be expected to be well-behaved. Even when driving the Hales Signatures' low impedance, I noted in the auditioning that "I was quite surprised by the 140's ability to produce fairly high playback levels before the sense of strain set in." With the right load and a judicious hand on the volume control, the 140 can be expected to deliver the musical sound I experienced during most of the auditioning.

Distortion at rated power with one channel driven (45W into 8 ohms and 70W into 4 ohms) was quite low at 0.016% and 0.026% respectively. The input voltage required to produce 1% THD (full power) at the output was 759mV, slightly lower than the specified 0.775V, which in itself is quite low. Although this low input sensitivity would indicate that the NAP 140 would be ideal with a passive control unit, Naim recommends against driving their power amplifiers with anything other than a Naim preamp. Again, a careful reading of the Naim warranty suggests that using a passive control unit will void the warranty.

The 140's voltage gain was 28.3dB, a fairly typical value. Output impedance measured 0.26 ohms at 20Hz and 0.27 ohms at 1kHz, rising to 0.3 ohms at 20kHz. This is fairly high for a solid-state amplifier. Channel separation (not shown) was 89dB at 1kHz, dropping to 72dB at 20kHz. The NAP 140 does not invert polarity. A low level of DC (8mV) was present at the loudspeaker outputs.

Finally, the severe distortion heard with bi-wired AudioQuest Green Hyperlitz was the result of the 140 becoming unstable due to its output not seeing enough inductance. As it doesn't have a series output inductor, the NAP 140 relies on the loudspeaker cable to provide the inductance necessary to define the ultimate HF rolloff. (Stray capacitances can otherwise turn the negative feedback into positive at ultrasonic frequencies, resulting in instability if the circuit still offers any gain at these frequencies.) Rather than keep loudspeaker cables as short as possible, Naim recommends a minimum of 10' per side of their own cable to ensure that the amplifier remains stable. Though 8' runs of flat cables like the Symo and AudioQuest F-14 posed no problems for the 140—flat, spaced-conductor cables have an inherently high inductance—one is cautioned against their use because of potential warranty problems.—Robert Harley