Columns Retired Columns & Blogs |
This is one of the best articles I have read by a Stereophile reviewer in a long time.
Mark
The measurements of the Allnic A-5000 DHT (serial no.24036) were performed with Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see the January 2008 "As We See It," and www.ap.com). The needle of the front-panel bias switch rested on the C of the Allnic logo, which is within the limits for output-tube bias current. I didn't need to adjust it, therefore.
The voltage gain into 8 ohms for single-ended sources was both on the low side for a power amplifier, and slightly lower than specified, at 23.7dB from the 16 ohm output-transformer tap, 23.1dB from the 8 ohm tap. For balanced sources, the gain was 4.5dB lower from both taps. The unbalanced input preserved absolute polarity, as did the balanced input when it was set to pin 2 hot with the rear-panel switch. The unbalanced input's input impedance met its 100k ohms specification, dropping inconsequentially to 59k ohms at 20kHz. The balanced input impedance was 3200 ohms across the audioband, which, along with the ability to invert the XLR's hot/cold wiring, suggests that this input is transformer coupled.
The A-5000's output impedance was impressively low for a single-ended tube design even from its 16 ohm tap, where it measured 1.6 ohms at 20Hz, 1.35 ohms at 1kHz, and 1.5 ohms at 20kHz. As expected, the respective impedances were lower from the 8 ohm tap: 1, 0.8, and 0.85 ohm. As a result, the modulation of the A-5000's frequency response that results from the Ohm's Law interaction between this output impedance and the impedance of the loudspeaker was low for a transformer-coupled tube amplifier, at ±0.6dB, 8 ohm tap (fig.1, gray trace); and ±0.9dB, 16 ohm tap (fig.2, gray trace). Figs. 1 and 2 also show that while there are well-damped ultrasonic resonances at 80 and 160kHz, these are better-suppressed when the load impedance is much lower than the nominal tap value. The same thing is true for the small rise in response below the audioband.
The presence of the ultrasonic resonances results in a small degree of overshoot in the shape of the amplifier's reproduction of a 1kHz squarewave (fig.3), but the flat top of this waveform correlates with the A-5000's well-extended low frequencies. The overshoot can also be seen in the Allnic's reproduction of a 10kHz squarewave (fig.4), but commendably, there is no ringing, confirming that the resonances are well damped.
Taken with the RCA input jack shorted, the A-5000's unweighted, wideband signal/noise ratio was good rather than great, at 64.9dB (ref. 2.83V into 8 ohms). As AD noted, this was primarily due to a small amount of hum at the AC supply frequency and its odd harmonics (fig.5), most likely due to magnetic coupling between the AC and output transformers. Switching an A-weighting filter into circuit improved the S/N ratio to 76.8dB.
Allnic specifies the A-5000 DHT's single 300B tube as producing a maximum power of 10W into 8 ohms (10dBW). Fig.6, taken from the 8 ohm tap, shows how the THD+noise percentage in the amplifier's output varied with power into 8 ohms. Our normal definition of clipping is when this percentage reaches 1%; fig.6 reveals that the Allnic delivered 4.7W at 1% THD (6.7dBW). However, this graph shows that the amplifier did reach its specified power at 1.8% THD, delivering 11.5W (13.6dBW) at the point where there is a "knee" in the trace. The distortion increased more rapidly into loads lower than the nominal tap value, with just 1.6W delivered into 4 ohms at 1% THD, and 6.5W into 4 ohms at 3% THD (fig.7). The picture was identical from the 16 ohm tap (fig.8), the specified power delivered into a matched load at just less than 2% THD, and much lower levels of power available into lower impedances.
Fig.9 plots how the THD+N percentage changes with frequency at a voltage equivalent to 500mW into 8 ohms. Other than at low frequencies, where magnetic saturation in the transformer core begins to have an effect, the distortion remains remarkably even with frequency, though it does rise alarmingly as the load impedance drops below the nominal tap value. With the 8 ohm tap driving 16 ohms, the distortion lies below 0.2% over much of the audioband. The 16 ohm tap behaved similarly, though with greater distortion overall (fig.10).
Fortunately for the A-5000's sound quality, the distortion is predominantly the innocuous second harmonic in nature (fig.11, and fig.12, blue trace), though the third, fourth, and fifth harmonics increase in level as the output current increases (fig.12, red trace). Though the Allnic A-5000 developed a second-order difference product at 1kHz at 50dB (0.3%) when reproducing an equal mix of 19 and 20kHz tones at 1W into 8 ohms (fig.13), higher-order intermodulation products were relatively low in level. Some low-level AC-supply modulation can also be seen in this graph.
Given the A-5000 DHT's single-ended topology, the presence of relatively high levels of second-order distortion in its output is inevitable. But the wide frequency response, excellent squarewave reproduction, and low output impedances, even from the 16 ohm output tap, are all commendable for such a design. One caution remains: If this amplifier is not to run out of steam, it must be used with speakers having an impedance equal to or greater than that of its nominal output transformer's tap.John Atkinson
This is one of the best articles I have read by a Stereophile reviewer in a long time.
Mark
Just having these amps in the house for a test run would impart a sense of joy.