Hovland Sapphire power amplifier Measurements
I carefully adjusted the Hovland Sapphire's bias before performing any measurements. This is more difficult than you might expect, due to some interaction between the settings. After you have adjusted the bias for the second tube in each pair, you need to readjust the first. The manual, however, is well-written and informative on the procedure.
The Sapphire's input impedance measured a high 68k ohms at 1kHz, this dropping to a still high 36k ohms at 20kHz. The amplifier didn't invert polarity with the red speaker lead connected to the Common terminal, as intended by Hovland. The voltage gain into 8 ohms was 27dB from the 16 ohm output-transformer tap, dropping to 25.5dB from the 8 ohm tap and a low 22.7dB from the 4 ohm tap.
The output impedance from the 16 ohm tap was a very high 3.55 ohms across most of the audioband, rising to 4.3 ohms at 20kHz. The picture looked a little better from the 8 ohm tap, at 2.25 and 2.6 ohms, respectively, while the source impedance from the 4 ohm tap was 1.4 ohms at 1kHz, rising to 1.55 ohms at 20kHz. Even so, there was significant modification of the amplifier's response due to the interaction between this highish source impedance and the manner in which the loudspeaker's impedance changes with frequency. This can be seen from the top dashed trace in fig.1, which reveals ±0.9dB response variations with Stereophile's standard simulated speaker load.
Fig.1 Hovland Sapphire, 4 ohm tap, frequency response at (from top to bottom at 2kHz): 2.83V into dummy loudspeaker load, 1W into 8 ohms, 2W into 4 ohms, 4W into 2 ohms (1dB/vertical div., right channel dashed).
Fig.1 also shows that the amplifier's ultrasonic response rolls off earlier with decreasing load impedance, though even in the worst case, driving 2 ohms from the 16 ohm tap, the response at 20kHz is down by only 0.6dB. However, you can see in this graph that a response peak develops between 100kHz and 200kHz, implying a parasitic resonance of some sort in this region. Fig.2 shows that this resonance adds a small degree of overshoot and ringing to the squarewave response. Its subjective consequences are hard to predict, but sources with relatively high amounts of ultrasonic energy, such as SACD, are best avoided with this amplifier.
Fig.2 Hovland Sapphire, 8 ohm tap, small-signal 10kHz squarewave into 8 ohms.
Channel separation at 1kHz (fig.3) is okay at 90dB (L-R) and 78dB (R-L), though the crosstalk increases with frequency at 6dB/octave, due to the usual capacitive coupling between the channels. Note, however, the peaks at 60Hz, 120Hz, and 180Hz in this graph, due to the presence of power-supply noise in both channels that I couldn't eliminate, no matter how I arranged the grounding in my test setup. Even though the highest in level of these peaks, at 60Hz, is still 74dB down, I still don't like to see hum like this in an amplifier's output. In an e-mail I received from Alex Crespi, Robert Hovland wrote, "We are aware that there is a very slight amount of measurable 60Hz hum component on the output line. However, the amplifier will meet its published S/N specs if the test criteria in the specs are followed. In our opinion, reducing this hum further will negatively affect the attractive musical qualities of the Sapphire."
Fig.3 Hovland Sapphire, channel separation (10dB/vertical div., R-L dashed).
Figs.4-6 show the Sapphire's small-signal THD+noise percentage plotted against frequency from the 16, 8, and 4 ohm output taps, respectively. Points to note from these graphs are that the Hovland amplifier is not happy driving impedances much below the nominal output-transformer tap rating, and that the distortion drops below 0.1% only with the 4 ohm tap driving 8 ohms. Note also the rise in THD in the mid- and low bass. The right channel, however, is a little more linear than the left, which might suggest that I'd set its output stage bias a little more optimally. (The bias meter was centered for all tubes.)
Fig.4 Hovland Sapphire, 4 ohm tap, THD+noise (%) vs frequency at (from top to bottom at 10kHz): 4W into 2 ohms, 2W into 4 ohms, 2.83V into simulated loudspeaker load, 1W into 8 ohms (right channel dashed).
Fig.5 Hovland Sapphire, 8 ohm tap, THD+noise (%) vs frequency at (from top to bottom at 10kHz): 4W into 2 ohms, 2W into 4 ohms, 2.83V into simulated loudspeaker load, 1W into 8 ohms (right channel dashed).
Fig.6 Hovland Sapphire, 16 ohm tap, THD+noise (%) vs frequency at (from top to bottom at 10kHz): 4W into 2 ohms, 2W into 4 ohms, 2.83V into simulated loudspeaker load, 1W into 8 ohms (right channel dashed).