Wavelength Audio Gemini monoblock power amplifier Measurements

Sidebar 3: Measurements

Measuring the Wavelength Audio Gemini took rather longer than I had anticipated, given the choice of output-stage tube and the two transformer taps. To keep things simple, I decided to measure the amplifier using just one example of each output tube: a Sylvania 45, which Robert Deutsch had found to be quietest; and a Sovtek 2A3, which RD had preferred to the NOS Raytheon and the microphonic AVVT mesh-plate 2A3.

There was also the issue of the tube-heater hum having to be manually nulled. I did my best, but the orientation of the chassis and the presence of the trimming tool each slightly affected the level of the hum, meaning that the precise null point shifted when the tool was no longer touching the potentiometer and the amplifier was set upright. The results regarding noise are therefore somewhat dependent on the precision with which I managed to null the hum: short-circuiting the input gave a hum figure of -66.5dB ref. 1W into 8 ohms with the 45 tube, -64.1dB with the 2A3.

(Commentary: In these days of high-precision, low-impedance, solid-state DC voltage regulators, I see no reason for Gordon Rankin using an AC heater supply other than the philosophical: to avoid having to use any solid-state devices in the Gemini. And even if it is felt that an AC supply sounds better, these levels of heater-injected hum are quite audible with the high-sensitivity speakers with which the Gemini is likely to be used.)

Getting the basics out of the way, the Wavelength's input impedance at 1kHz was a sensible 46.3k ohms, and the amplifier preserved absolute polarity. Its voltage gain into 8 ohms was very much lower than the usual 27dB, ±3dB we see these days, and was dependent on both the tube and the transformer tap: with the 2A3 tube, the gain at 1kHz was 13.4dB and 11.2dB from the 8 and 4 ohm taps, respectively; with the 45 tube, the respective figures were 10.5dB and 9.1dB. Regardless of the output tube, the amplifier clipped (3% distortion) with an input of 1.32V (4 ohm tap) or 1.2V (8 ohm tap), meaning that a typical CD player should not be fed straight into the Gemini without at least 6dB of attenuation.

The source impedance was on the high side and depended on the output tube, the output transformer tap, and the signal frequency. With the 2A3 tube, it also varied inversely with the current drawn from the tube, decreasing by about 17% when the load dropped from 8 to 4 ohms. The low-frequency source impedance with the 2A3 was 3.7 ohms (8 ohm tap) or 1.7 ohms (4 ohm tap); the midrange impedance was 3.1 and 1.5 ohms, respectively; while at 20kHz, the impedance rose to 4.6 and 1.8 ohms, respectively. The respective figures with the 45 tube were 3.2 and 1.5 ohms, 5.5 and 2.6 ohms, and 6.8 and 2.9 ohms.

As a result, there will be considerable modulation of the Gemini's frequency response due to the interaction between its source impedance and the loudspeaker's load impedance, particularly with the 45 tube. This can be seen in figs.1 and 2, which show the amplitude response of the 8 ohm output with the 45 tube (fig.1) and the 2A3 (fig.2). The response with Stereophile's simulated speaker load changes by almost ±3dB with the 45 tube, ±2dB with the 2A3. With the 4 ohm output transformer tap the variations drop to ±2dB and ±1dB, respectively (not shown). Into resistive loads, the low bass rolls off to -3dB at 24Hz (45) or 18Hz (2A3), the 2A3 rolloff showing a sharper "knee." The high-frequency responses drop by a small fraction of a dB at 20kHz, with the 45 traces showing a less-well-damped ultrasonic resonance around 120kHz, presumably due to its higher internal impedance.

Fig.1 Wavelength Gemini, 45 tube, 8 ohm tap, frequency response at 1V into (from top to bottom at 2kHz): dummy loudspeaker load, 16 ohms, 8 ohms, 4 ohms, and 2 ohms (2dB/vertical div.).

Fig.2 Wavelength Gemini, 2A3 tube, 8 ohm tap, frequency response at 1.4V into (from top to bottom at 2kHz): dummy loudspeaker load, 8 ohms, 4 ohms, and 2 ohms (2dB/vertical div.).

The 1kHz squarewave responses (fig.3, 45; fig.4, 2A3) feature an excellent shape, with the increased slope-back on the 45 tube's response indicating its earlier low-bass rolloff. (As the Avantgarde horns used by RD for his auditioning use a powered subwoofer, this would not have been a factor in his preference for the 45 tube.) The Gemini's reproduction of a 10kHz squarewave (fig.5, 45) was also excellent.

Fig.3 Wavelength Gemini, 45 tube, 8 ohm tap, small-signal 1kHz squarewave into 8 ohms.

Fig.4 Wavelength Gemini, 2A3 tube, 8 ohm tap, small-signal 1kHz squarewave into 8 ohms.

Fig.5 Wavelength Gemini, 45 tube, 8 ohm tap, small-signal 10kHz squarewave into 8 ohms.