Integrated Amp Reviews

I performed the measurements of the Fezz Audio Silver Luna Prestige amplifier with my Audio Precision SYS2722 system, confirming some of the results with the higher-performance APx500. I carefully installed the tubes, following the numbered guide that had been packed with the amplifier. I carried out a complete set of tests with the 12AX7 tubes that AH had told me he preferred, with the output stage in both pentode and tetrode modes. I then repeated some of the testing with the alternative 6N2P small-signal tubes. I powered down the amplifier before changing the output modes or operating the input tube switch on the top panel. The 6N2P is not pin-compatible with the 12AX7; I assume the switch changes the internal connections.

As usual, I waited for an hour after powering up the amplifier before starting the testing. The Silver Luna Prestige preserved polarity from both the 4 ohm and 8 ohm transformer taps in both output modes. The input impedance with the 12AX7 tubes was a usefully high 74k ohms at 20Hz and 1kHz, 45k ohms at 20kHz. The input impedance was slightly lower with the 6N2P tubes, but still high at 69k ohms at 20Hz, 67k ohms at 1kHz, and 45k ohms at 20kHz. Both channels' source impedance from the 8 ohm and 4 ohm output transformer taps was the same in both the pentode and tetrode modes. From the 8 ohm taps, it was a very high 6.5 ohms at 20Hz, 5.4 ohms at 1kHz, and 6.25 ohms at 20kHz. From the 4 ohm taps, the source impedance was a still-high 3.2 ohms at 20Hz, 2.8 ohms at 1kHz, and 3 ohms at 20kHz.

When I estimated the voltage gain at 1kHz, I found a significant difference between the two channels. With the volume control set to its maximum and the 12AX7 tubes, the left channel's gain in pentode mode into 8 ohms was 31.6dB from the 8 ohm tap and 30.52dB from the 4 ohm tap. The right channel's maximum gain was almost 3dB lower from both taps, however. The differences in gain were the same at lower settings of the volume control, in tetrode mode, with the 6N2P tubes, and with the APx500 analyzer. I tried swapping the 12AX7 tubes between the channels, but the right channel's gain was still lower than the left's.

This difference in channel gains can be seen in fig.1, which plots the Fezz's small-signal frequency response from the 8 ohm tap in pentode mode. The response in both channels is flat to well above the audioband, with no resonant peaks present, but starts to roll off in the low bass. The amplifier's high source impedance means the variation in the frequency response with our standard simulated loudspeaker (gray trace) was high, at ±3.4dB. The response variation with the simulated speaker from the 4 ohm tap was lower, at ±2,1dB. However, there will be audible modifications of loudspeaker responses with both output taps. The Fezz's reproduction of 1kHz and 10kHz squarewaves into 8 ohms (figs.2 & 3) was excellent, with no overshoot or ringing.

The Silver Luna Prestige's channel separation (not shown) was good below 3kHz, at >70dB, L–R, and >80dB, R–L. The separation decreased by 15dB in both directions at the top of the audioband, but this is still good performance. In pentode mode with the 12AX7 tubes, the unweighted, wideband signal/noise ratio (ref. 1W into 8 ohms), taken from the 8 ohm taps with the inputs shorted to ground and the volume control set to its maximum, was 62.9dB in the left channel, 66.9dB in the right. These ratios improved slightly when the measurement bandwidth was restricted to 22Hz–22kHz, and to 75.0dB, left, and 81.4dB, right, when A-weighted. With their lower gain, the ratios from the 4 ohm outputs were 1dB higher and another 1dB higher from both taps in tetrode mode.

Spectral analysis of the low-frequency noisefloor while the Fezz's 8 ohm taps in tetrode mode drove a 1kHz tone at 1Wpc into 8 ohms, with the volume control set to the maximum (fig.4), revealed a low level of random noise. However, AC supply–related spuriae at 60Hz and its odd- and even-order harmonics were present in both channels. The highest of these spuriae, at 60Hz and 120Hz, lay close to –73dB (0.02%). Repeating the analysis with the volume control set to –20dB (and the input signal increased by 20dB) didn't change the levels of these spuriae.

Fezz specifies the Silver Luna Prestige's maximum power as 35Wpc, which is equivalent to 15.44dBW into 8 ohms and 12.44dBW into 4 ohms. Looking at how the left channel's THD+N percentage changed with power, the Silver Luna Prestige didn't meet its specified power with our usual definition of clipping, which is when the THD+noise reaches 1%. Significantly relaxing that definition to 10% THD+N, the amplifier did meet its specified power in pentode mode from both the 8 ohm tap into 8 ohms (fig.5) and from the 4 ohm tap into 4 ohms (fig.6). Less power was available in tetrode mode, the THD+N from the left channel's 8 ohm tap reaching 10% at 29W (14.62dBW, fig.7) and 23.4W into 4 ohms (10.68dBW). The rise in THD+N as the power increases above 500mW in these graphs suggests that the circuit has little or no loop negative feedback.

When I examined how the percentage of THD+N in both channels varied with frequency, I found that not only did the right channel have lower gain than the left; it also had a much higher level of distortion. This can be seen in fig.8, taken with the 4 ohm tap in pentode mode at 2.83V into 8 ohms (left channel blue, right red) and 4 ohms (left green, right gray). Something was obviously wrong in the right channel's circuitry—I therefore continued the testing using just the left channel.

The distortion waveform in pentode mode was predominantly the subjectively innocuous second harmonic (fig.9), while the third harmonic made an appearance in tetrode mode (fig.10). While higher-order harmonics are present, these are much lower in level in pentode mode (fig.11), but not in tetrode mode (fig.12).

With the Silver Luna Prestige's 8 ohm tap in tetrode mode driving an equal mix of 19 and 20kHz tones at 1W peak into 8 ohms (fig.13), the 1kHz difference product lay at –56dB (0.15%), though the higher-order intermodulation products were much lower in level. The noisefloor was relatively high, however.

The Fezz Silver Luna Prestige's measured performance is a mixed bag. The cleanly extended high frequencies and excellent squarewave performance, coupled with the fact that the distortion doesn't rise at low frequencies, indicate that the output transformers are of high quality. But the high source impedance and the relatively high level of distortion, even in the left channel, suggest that the amplifier will sound different with every loudspeaker with which it is paired.

As for the lower gain and much higher distortion in the right channel, these suggest that something was broken in that channel's circuitry, presumably after Alex completed his audition, as a 3dB intra-channel difference would be very audible. The amplifier's packaging was first-class, so I very much doubt that the damage occurred in shipping. I initially wondered if the problem was due to the input tube switch malfunctioning. However, the different behavior in the two channels was consistent with both tube types, which rules that out. All I can think of is that the sockets for the input tubes were very tight, which required more force than I was expecting to properly seat them. If these sockets are mounted on a printed circuit board, it is possible that repeated removal and insertion of the tubes broke a circuit trace.—John Atkinson