Test Report: Bryston SP-3 Preamplifier/Processor and 9BSST2 5-Channel Amplifier Page 3

Bottom Line

I am deliberately stating the obvious when I observe that for what you’d pay for an SP-3/ 9BSST2 combo, you might well purchase a modest, foreclosed-and-abandoned home in certain neighborhoods in, say, Florida or Las Vegas. But that’s hardly the point — nor is the news that a buyer even mildly curious about the price would not be shopping in these environs in the first place.

For the 1 percent left standing (more like 0.1 percent, I’ll warrant), there’s no argument: Bryston’s surround electronics are state-of-the-art. They are as well made as any consumer electronics you can buy, and in a system designed, installed, and programmed for “just press play” operability, with loudspeakers and room acoustics of concomitant quality, audio playback simply ain’t gonna get any better. Which is what that oft-abused phrase “state of the art” is supposed to mean.

Test Bench

Note: All tests performed on SP-3/9BSST2 together, connected via unbalanced inputs/outputs.

All data were obtained from various test DVDs using 16-bit dithered test signals, which set limits on measured distortion and noise performance. Reference input level is –20 dBFS, and reference output is 1 watt into 8 ohms. Volume setting for reference level was -3.5. All level trims at zero; except for subwoofer-related tests, all speakers were set to “large,” subwoofer on. All are worst-case figures where applicable.

Output at clipping (1 kHz into 8/4 ohms)

  • 1 channel driven: 182/247 W (22.6/23.9 dBW)
  • 5 channels driven (8 ohms): 131 W (21.2 dBW)

Distortion at 1 watt (THD+N, 1 kHz)

  • 8/4 ohms: 0.02/0.02%

Noise level (A-wtd): –75.0 dB

Excess noise (with sine tone)

  • 16-bit (EN16): 1.9 dB

Frequency response: 20 Hz to 20 kHz +0, –0.4 dB

Reference input and output level is 200 mV; volume setting for reference output level was –3.5.

  • Distortion (THD+N, 1 kHz, 8 ohms): 0.02%
  • Noise level (A-wtd): –86.3
  • Frequency response: <10 Hz to 71 kHz +0, –3 dB

Reference level is –20 dBFS; all level trims at zero. Volume setting for reference level was –3.

  • Output at clipping (1 kHz, 8/4 ohms, both channels driven): 148/226 W (21.7 /23.5 dBW)
  • Distortion at reference level: 0.02%
  • Linearity error (at –90 dBFS): see notes
  • Noise level (A-wtd): –74.7 dB

with 96-kHz/24-bit signals: –79.4 dB

  • Excess noise (with/without sine tone)

16-bit (EN16): 1.6/1.2 dB

quasi-20-bit (EN20): 13.9/15.2 dB

  • Noise modulation: 1.6 dB
  • Frequency response: <10 Hz to 20 kHz +0, –0.4 dB

with 96-kHz/24-bit signals: <10 Hz to 44 kHz +0, –2 dB

Measured results obtained with Dolby Digital test signals.

  • Subwoofer-output frequency response (crossover set to 80 Hz): 24 dB/octave above –6-dB rolloff point of 80 Hz
  • High-pass-filter frequency response (crossover set to 80 Hz): 12 dB/octave below –3-dB rolloff point of 80 Hz
  • Maximum unclipped subwoofer output (trim at 0): 5v (approximately; see notes)
  • Subwoofer distortion (from 6-channel, 30-Hz, 0-dBFS signal; subwoofer trim set to 0): see notes
  • Crossover consistency: bass crossover frequency and slope were consistent for all sources and formats.
  • Speaker-size selection: all channels can be set to “small”
  • Speaker-distance compensation: available for all main channels.

Bryston’s multichannel SP-3 and 9BSST2 pairing produced varying results on the test bench, which, a bit paradoxically, is fairly typical of high-end designs. Amplifier output was impressive: The 9BSST2 easily beat its highly conservative 120-watts-per-channel rating, even with all five channels driven. Interestingly, the Bryston amp yielded only modestly more watts driving 4-ohm loads than with 8 ohms, suggesting that its design features substantially more of an output-device safe operating area than “necessary” for its overall current potential; this might have some bearing on the outstanding dynamic freedom I heard.

Frequency response tests uniformly reflected a gentle rolloff (about 3 dB/octave) over the ultrasonic octaves, beginning as a 0.4-dB deficit at 20 kHz. Noise performance was uniformly good: in digital-signal tests, about 1 dB short of the theoretical limits using S+V’s real-world dithered signals. S/N on 96/24 PCM enjoyed an advantage of only about 5 dB, suggesting analog-circuit noise as the limiting factor. (While generalization is always dangerous, I have found this kind of S/N result fairly usual among high-end, “all-discrete-component” designs, whose creators accept the slight noise penalty in exchange for other subjective sonic virtues they value more highly.)

Two head-scratchers: First, while the SP-3’s digital-to-analog linearity with dithered tones was essentially perfect to –80 dBFS, at –90 dBFS its output went vastly negative (–120 dBFS or below), suggesting that its D/A system was muting, or making a many-decibel calculation error on this particular tone. However, because I found response on a –100 dBFS tone (which we do not regularly report, as it pushes the boundary of our measurement repeatability) to be quite close, and because, despite many efforts on headphones, I could not hear any sign of D/A discontinuity on our dithered continuous “fade-to-noise” listening-test signals, I consider this more of a test-bench anomaly than a performance problem.

Second, the SP-3 could not pass our subwoofer-output torture test (full-scale 31-Hz signal present in all channels) without digi-clipping it quite severely, and the same was true of 5 x 30-Hz 0-dBFS, and even 6 x 30-Hz -3-dBFS signals from the Dolby Digital test disc. (Both these signals have issues that disqualify them for single-test sub-out evaluations, but I ain’t going into that here…) Yet sub output with “normal” signals (–20 dBFS-ish) was beautiful (0.02 % THD), and the output could deliver around 5 volts, which is plenty to drive any sub system to full output. So, with very few reservations, I class this as another non-problem, because full-scale signals at low frequencies are rare even in one or two channels at a time, let alone five or six. And still more so because the audible impact is questionable anyway, since the distortion products mostly arise beyond the passband of the low-pass filter.