Integrated Amp Reviews

Though no larger than a paperback book, the PS Audio Sprout is a complex product, with USB, S/PDIF, and Bluetooth digital, analog line-level, and analog phono inputs, and speaker outputs, a headphone output, and a line-level output. Testing the Sprout was therefore more complicated than with a typical integrated amplifier. I measured the Sprout using my recalibrated Audio Precision SYS2722 system (see www.ap.com, and the January 2008 "As We See It"). As the Sprout is a class-D amplifier—it uses an output module sourced from the Scandinavian Elan/Abletec company—it produces ultrasonic noise that would overload the Audio Precision's input circuitry. I therefore performed most of the tests using, ahead of the analyzer, an Audio Precision AUX-0025 passive low-pass filter, which eliminates noise above 200kHz. (Without the filter and with no signal, there was 200mV of ultrasonic noise with a center frequency around 400kHz present at the Sprout's speaker terminals.)

Looking first at the PS Audio Sprout as a conventional integrated amplifier, and driving its line input, it offered a maximum voltage gain of 25.8dB into 8 ohms and preserved absolute polarity (ie, was non-inverting). The input impedance was on the low side, at 8.25k ohms at high and middle frequencies, rising slightly to 9.4k ohms at the bottom of the audioband. The output impedance was extremely low, at 0.03–0.05 ohm, including 6' of speaker cable.

As a result, the variation in response with our standard simulated loudspeaker (fig.1, gray trace) was negligible, nor was there any change in response as the load dropped from 8 to 2 ohms. However, note the 7.5dB peak centered on 67Hz in this graph. It appears that the Sprout's speaker outputs offer some compensation for the rolled-off low frequencies typical of small, inexpensive speakers, then quickly roll off the low bass to avoid overload. This equalization would work well with Herb Reichert's KEF and Totem speakers, but should have made his full-range Tekton Enzos sound too ripe. However, it is fair to note that he didn't comment on any such ripeness. Fig.1 was taken with the volume control set to its maximum. Repeating the measurement with it set to 12 o'clock gave a slightly lower bass peak but preserved the extended high-treble response. The curtailed low bass correlates with the sloped tops and bottoms of the waveform with a 1kHz squarewave (fig.2), and a slight overshoot is visible on the leading edges. However, looking at a 10kHz squarewave (fig.3) reveals that this overshoot is not associated with any ringing.

Channel separation was good rather than great, at 65dB in both directions at 1kHz, this decreasing to 40dB at 20kHz. With the Audio Precision low-pass filter in circuit, the Sprout's wideband, unweighted signal/noise ratio was modest, at 57.3dB left and 55.2dB right, though restricting the measurement bandwidth to the audioband respectively improved these ratios to 64.9 and 61.4dB; A-weighting the measurement gave further improvement, to 80 and 82dB. Spectral analysis of the amplifier's low-frequency noise floor while it reproduced a 1kHz signal at 1W into 8 ohms (fig.4) revealed that there was a regular series of power-supply–related spuriae in the Sprout's output, with the 60Hz component higher in the right channel than the left. But these spuriae are still sufficiently low not to be heard as hum.

The Sprout is specified as having maximum power outputs of 33W into 8 ohms (15.1dBW) and 50W into 4 ohms (14.0dBW). With clipping defined as the power when the THD+noise in the amplifier's output reaches 1%, the Sprout with both channels driven clipped at 32Wpc into 8 ohms (15.05dBW, fig.5) and 57Wpc into 4 ohms (14.55dBW, fig.6). The downward slope of the traces in each of these graphs indicates that the Sprout's distortion actually lies beneath the noise up to 20W or so, and is very low.

Usually, I would have plotted how the percentage of THD+N varied with frequency at the voltage where the distortion started to rise above the noise floor. However, because the bass boost will reduce the Sprout's maximum output at low frequencies, I examined this at 3V output. Fig.7 reveals that the THD doesn't change with frequency other than in the very low bass, where the rise is actually associated with the response rolloff. At 1kHz, the distortion appears to be predominantly second harmonic (fig.8), though I had to average 64 captures to bring the distortion waveform out of the noise that would otherwise obscure it. I performed spectral analysis of the amplifier's output with a 50Hz tone into 4 ohms (fig.9) at a lower power than usual, because the Sprout's protection circuit operated at higher continuous levels into this impedance. But again the second harmonic is the highest in level, though actually lower in level than the 120Hz supply-related tone.

Tested for intermodulation distortion with a 1kHz-spaced pair of high-frequency tones (fig.10), the Sprout produced a 1kHz difference tone at an acceptably low –63dB (0.07%), though the higher-order products at 18 and 21kHz were 12dB higher in level.

The headphone output (which mutes the speaker outputs), again tested with a line-level input signal, offered a maximum gain of 12.4dB and, again, was non-inverting, as was the output at the line output. (This output offered a maximum gain of 6.4dB.) The headphone output offered a very low source impedance of less than 1 ohm, while the line output jack offered 235 ohms. (Both values applied at all audio frequencies.) These outputs, taken ahead of the main amplifier's class-D stage, behaved differently from the speaker outputs in the high treble and bass. Fig.11 shows the frequency response at the headphone output with the volume control set to its maximum (blue and red traces) and to 12 o'clock (cyan, magenta). The response rolls off above the audioband, and has no boost in the midbass at the lower control setting, and only a hint of boost at the maximum setting. However, the low bass is still rolled off below 70Hz, which is perhaps why HR found the Sprout to sound dry and lean with headphones compared with his reference. Harmonic distortion from the headphone output was very low (fig.12), as was intermodulation distortion (not shown).

I measured the Sprout's phono-stage performance at the headphone output, so it came as no surprise to find that the response with RIAA correction and the volume control set to its maximum was very similar to the line input response (fig.13). The ultrasonic rolloff occurs earlier, however, and is –1.5dB at 20kHz. But note the accuracy of the RIAA equalization and the very close matching between the two channels. The phono stage was again non-inverting, and the maximum gain for phono sources was 53.3dB at the headphone output and 66.7dB at the speaker outputs, these appropriate for moving-magnet cartridges. The input impedance was also compatible with MM cartridges, at 42k ohms at 2Hz, 45k ohms at 1kHz, and 37.5k ohms at 20kHz.

The low-frequency phono-overload margin was excellent, at 20dB at 20Hz (ref. 1kHz at 5mV), though this will be associated with the rolled-off bass. The margin dropped to an adequate 11.4dB at 1kHz and 12.6dB at 20kHz. Distortion with a 1kHz tone at 5mV was low, with the second harmonic the highest in level at just –86dB (0.006%, fig.14).

Finally, I examined the performance of the Sprout's D/A converter with both USB and S/PDIF data. (I didn't test it with Bluetooth data, as its behavior will be dominated by the lossy codec used.) With the volume control set to its maximum, a –12dBFS signal at 1kHz gave rise to 10.03V at the speaker outputs, 2.14V at the headphone outputs. A full-scale digital signal would therefore clip both outputs with the volume control at its maximum; fortunately, the control will not be used at its maximum. Apple's USB Prober utility identified the Sprout as "XMOS USB 2.0 Audio Out," indicating that the USB receiver uses the popular XMOS device and that it operated in the optimal isochronous asynchronous mode.

The D/A section preserved absolute polarity, and its impulse response with 44.1kHz data (fig.15) indicates that the digital reconstruction filter is a conventional FIR type. Wideband spectral analysis at the headphone output with 44.1kHz-sampled white noise (fig.16, red and magenta traces) reveals that the filter has a fairly slow initial rolloff, which means that the ultrasonic image of a full-scale 19.1kHz tone at 25kHz is suppressed by 60dB rather than being buried in the stop-band noise floor. Both the USB and S/PDIF inputs locked to data of all sample rates from 44.1 to 192kHz. A more conventional look at the baseband frequency responses, with data sampled at 44.1, 96, and 192kHz (fig.17), shows that the ultrasonic rolloff is identical with all three rates, other than a sharp drop just below each Nyquist frequency (half the sample rate) for 44.1 and 96kHz data.

The drop in the noise floor that occurred when feeding the digital inputs first with 16-and then 24-bit data, both representing a dithered 1kHz tone at –90dBFS (fig.18), suggested that the Sprout offers about 17 bits of digital resolution for both USB and S/PDIF data. With an undithered tone at exactly –90.31dBFS, the Sprout successfully resolved the three DC voltage levels described by the digital data (not shown). Feeding the Sprout's digital inputs with the Miller-Dunn J-Test signal gave the same result (fig.19): too high a noise floor, which completely obscures the odd-order harmonics of the low-frequency, LSB-level squarewave; and a regular series of sidebands of unknown origin, spaced at ±800, ±1600, ±2400Hz, etc.

Though the Sprout didn't excel in any area of measured performance, it compensates for that with its affordable price and considerable flexibility. It really is a "one-stop shop" for putting together an inexpensive system.—John Atkinson