PS Audio HCA-2 power amplifier John Atkinson January 2006
When Kalman Rubinson reviewed PS Audio's HCA-2 for the December 2002 issue (Vol.25 No.12), he enthused over the sound of this 150Wpc, $1695 power amplifier, but I was bothered by some anomalous measured performance. The amplifier was less linear, particularly at high frequencies, than I had expected. Fig.1, for example, shows how the amplifier's THD+noise percentage varied with frequency at 2.83V into 8, 4, and 2 ohms. Acceptably low in level into the higher impedances and at midrange and low frequencies, and predominantly the subjectively benign third harmonic (fig.2), the measured THD+N increased significantly above 1kHz and into lower impedances, due to the introduction of higher-order distortion harmonics (not shown).
Fig.1 PS Audio HCA-2, THD+N (%)vs frequency at 2.83V into (from bottom to top): 8, 4, 2 ohms.
Fig.2 PS Audio HCA-2, 1kHz waveform at 1W into 8 ohms (top), 0.16% THD+N; distortion and noise waveform with fundamental notched out (bottom, not to scale).
Was this really how the amplifier measured? Admittedly, the HCA-2 is a low-feedback amplifier, a type of design that will have higher levels of static distortion than a high-feedback amplifier—but PSA's Paul McGowan had thought this an acceptable tradeoff. However, the HCA-2 also uses a class-D output stage, which inherently produces ultrasonic switching noise. It was conjectured that the measured distortion was actually due not to the HCA-2 but to the slew-rate limiting, due to the presence of this noise, of the input stage of my Audio Precision System One test set.
Audio Precision has published a white paper on this subject, "Measuring Switch-mode Power Amplifiers." "Within the current state of the art," wrote author Bruce Hofer, "it is simply not possible to design a high impedance analog input stage having audio-worthy distortion and noise performance without incurring a practical slew rate limitation." Audio Precision makes available a precision passive low-pass filter for measuring switching amplifiers.
Paul McGowan sent me measurements he had made of an HCA-2's THD+noise against frequency both using the Audio Precision filter and feeding the amplifier straight into the analyzer's input. His results are shown in fig.3. Without the AP filter (blue trace), the noise dominates the measured percentage. With the filter (red trace), the true THD is revealed as being very low in level until higher frequencies, when it rises.
Fig.3 PS Audio HCA-2, THD+N (%)vs frequency at 2.83V into 8 ohms with (red) and without (blue) Audio Precision low-pass filter (PS Audio measurement).
I hung on to the review sample of the HCA-2, intending to remeasure it using an audioband low-pass filter between its output and the audio analyzer, to eliminate the effect of the switching noise. Unfortunately, as with all things on my "to do" list that are categorized as "Important but Not Urgent," this intention remained unrealized for way too long. It was only when Robert Deutsch and I were discussing his review of PS Audio's GCC-100 integrated amplifier (see this issue's Equipment Reports) that I remembered the unfinished task.
In the past couple of years, when measuring switching amps, I've used an active sixth-order low-pass filter made by Tripath. Paul McGowan suggested that, when he sent me a sample of the GCC-100 for measurement, he also send me his Audio Precision low-pass filter, which would allow high-power measurements to be made. This was an excellent suggestion. After I had finished measuring the performance of the GCC-100, I retrieved the HCA-2 from storage and set it up on my test bench.
Without the AP filter and despite its long rest, the HCA-2 measured identically to how it had done three years before. I then hooked up the low-pass filter as recommended in the AP white paper and repeated the THD+N vs frequency and waveform measurements that had produced figs.1 and 2. The results are shown in figs.4 and 5. Despite the Audio Precision System One being shielded from the ultrasonic noise, both the measured levels of THD+N in the PS Audio's output and the character of the distortion spuriae were identical. (To check that I was using the AP filter correctly, I repeated the measurements using the active Tripath filter. The results were again identical.)
Fig.4 PS Audio HCA-2, THD+N (%)vs frequency at 2.83V into (from bottom to top): 8, 4, 2 ohms; with Audio Precision low-pass filter.
Fig.5 PS Audio HCA-2, 1kHz waveform at 1W into 8 ohms (top), 0.16% THD+N; distortion and noise waveform with fundamental notched out (bottom, not to scale); with Audio Precision low-pass filter.
While the problems of measuring switching amplifiers are real, in the case of the HCA-2 it appears that I had correctly characterized its distortion performance in the original review. The switching noise present in its output, which was at a much lower level than with the PSA GCC-100 or with the Yamaha MX-D1 reviewed last April by Michael Fremer, had not induced slew-rate limiting in my System One's input stage—or if it had, had not done so to the extent that it had corrupted the measurement.
What does this mean? Given the disparity between my measurements with the AP filter and PS Audio's, it is possible that our review sample was damaged in some way.
Both Kal Rubinson and Sam Tellig (in his October 2002 "Sam's Space") enjoyed the sound of the HCA-2. This suggests that, at the small-signal power levels that are typical of normal music listening, and with the power spectrum of the classical music that Kal and Sam predominantly listen to, the HCA-2's diminished linearity at high frequencies and into low impedances is much less of an issue than it would be listening to rock music at high levels. But as you can see from the measurements of the GCC-100 elsewhere in this issue, the current-generation PS Audio switching amplifiers offer considerably better measured performance.—John Atkinson