Wavelength Audio Gemini monoblock power amplifier Measurements part 3

Despite the amplifier's "bent" transfer function, levels of intermodulation distortion were lower than I expected. As can be seen in figs.12 and 13 (45 and 2A3, respectively), while the 1kHz difference component with either tube reached -50dB (0.3%), the higher-order spuriae lay below -60dB (0.1%). The furriness of the FFT-derived traces is due to the presence of hum in the Gemini's output signal.

Fig.12 Wavelength Gemini, 45 tube, 8 ohm tap, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 2V into 8 ohms (linear frequency scale).

Fig.13 Wavelength Gemini, 2A3 tube, 8 ohm tap, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 2V into 8 ohms (linear frequency scale).

Note the very low power levels used to derive the preceding graphs. This is because the Gemini really can't put out much in the way of power. Figs.14-17 show the THD+noise percentage plotted against output power in watts for the 45 and 2A3 tubes, from the 4 and 8 ohm taps, respectively. Again, the importance of matching the transformer tap to the load can be seen, as can the extra power available from the 2A3 tube. At 3% THD with the 45 tube, we're talking about 2.1W into 8 ohms (8 ohm tap) and 2.3W into 4 ohms (4 ohm tap). With the 2A3 tube, the respective figures are 3.8W and 4.1W!~ But in all cases the available power drops off sharply into higher and lower impedances, which can also be seen in the plots of output power against distortion with a low-duty-cycle 1kHz toneburst (fig.18, 45; fig.19, 2A3). Note that these curves show that, other than into 16 ohms, the amplifier is nowhere near actual clipping despite the high levels of distortion. Particularly into the lower impedances, the bent transfer function merely results in almost linear increases in distortion with increasing output power: classic "soft" clipping behavior.

Fig.14 Wavelength Gemini, 45 tube, 4 ohm tap, distortion (%) vs continuous output power (W) into (from bottom to top at 100mW): 4 ohms, 8 ohms, 16 ohms, and 2 ohms.

Fig.15 Wavelength Gemini, 45 tube, 8 ohm tap, distortion (%) vs continuous output power (W) into (from bottom to top at 100mW): 8 ohms, 4 ohms, 16 ohms, and 2 ohms.

Fig.16 Wavelength Gemini, 2A3 tube, 4 ohm tap, distortion (%) vs continuous output power (W) into (from bottom to top at 100mW): 8 ohms, 4 ohms, 16 ohms, and 2 ohms.

Fig.17 Wavelength Gemini, 2A3 tube, 8 ohm tap, distortion (%) vs continuous output power (W) into (from bottom to top at 100mW): 8 ohms, 4 ohms, 16 ohms, and 2 ohms.

Fig.18 Wavelength Gemini, 45 tube, 8 ohm tap, distortion (%) vs 1kHz burst output power into 16 ohms (blue trace), 8 ohms (green), 4 ohms (magenta), 2 ohms (red), and 1 ohm (black).

Fig.19 Wavelength Gemini, 2A3 tube, 8 ohm tap, distortion (%) vs 1kHz burst output power into 16 ohms (blue trace), 8 ohms (green), 4 ohms (magenta), 2 ohms (red), and 1 ohm (black).

This measured performance looks worse than it sounds, given that RD was probably not asking for more than a couple of hundred milliwatts from the Gemini during his auditioning with the Avantgarde horns; at his preferred levels, the amplifier was probably behaving in quite a linear manner. However, I still would like to have seen less hum in the Gemini's output. And I am still bothered by the "fix" that had to be applied to the amplifier's grounding by Neil Muncy.—John Atkinson

X