Legend Audio Design Starlet integrated amplifier Measurements part 2

Legend states in its literature that the Starlet's measured performance will be worse than that of a "$100 RadioShack amplifier." This was certainly the case when I examined the amplifier's levels of distortion. The literature implies that the Starlet's higher-than-usual THD number was deliberate and "contributes to the perception of power." But when you look at fig.3, which shows the percentage of THD+noise plotted against frequency at 2.83V into a variety of loads, including a simulated loudspeaker, it is difficult, for me at least, to sympathize with this design approach. Levels of THD between 3% and 0.6% in the low frequencies may sound more "powerful"—I note that CS did find the Starlet's bass to have "realistic weight and presence"—but in my opinion, it's a trick.

Fig.3 Legend Starlet, THD+noise (%) vs frequency at (from top to bottom at 4kHz): 4W into 2 ohms, 2.83V into simulated loudspeaker load, 2W into 4 ohms, and 1W into 8 ohms.

Spectral analysis of a high-level, low-frequency tone (fig.4) reveals that the second harmonic is the highest in level, with then a regularly decreasing series of higher distortion harmonics. AC supply components can also be seen at 60Hz and its odd harmonics—the former actually lies at -50dB (0.3%), which is pretty poor engineering. (I tried all manner of grounding arrangements to reduce the level of hum, to no avail.) Higher in frequency and at lower signal levels, the distortion shifts in favor of the third harmonic (fig.5). But at high levels, the second again becomes predominant (fig.6). (Note also in this graph the AC-supply-induced sidebands around each of the signal's spectral components.)