NuForce Icon USB-input integrated amplifier Measurements
I measured the NuForce Icon using Audio Precision's top-of-the-line SYS2722 system (see the January 2008 "As We See It" and www.ap.com), an Audio Precision System One Dual Domain, and the Miller Audio Research Jitter Analyzer.
Looking first at the Icon's performance as an amplifier, it offered a low maximum voltage gain of 22.6dB into 8 ohms from its speaker outputs, 6dB from its line-output jack. The line and headphone outputs preserved absolute polarity (ie, were non-inverting), while the speaker outputs inverted polarity. The input impedance was a reasonably high 46k ohms at 1kHz, dropping slightly at the frequency extremes. The line jack's output impedance was a usefully low 200 ohms in the midrange and 75 ohms at 20kHz, but rose to a very high 20k ohms at 20Hz. If the NuForce is used as a preamplifier, the power amplifier needs to have an input impedance of at least 100k ohms if the lows are not to be prematurely rolled off. The headphone jack offered a maximum gain of 15.1dB and an output impedance of 48 ohms across the audioband, which will make the Icon more suitable for use with high-impedance headphones like the Sennheisers rather than the low-impedance Grados and the Phiaton MS400.
The Icon's output impedance at the speaker jacks was a fairly low 0.13 ohm at low and midrange frequencies, rising inconsequentially to 0.14 ohm at 20kHz. As a result, the modification of the amplifier's frequency response by the usual Ohm's Law interaction between this source impedance and the impedance of our standard simulated loudspeaker remained within tight ±0.1dB limits (fig.1, black trace). This graph indicates that the 3dB points lie at 15Hz and 96kHz, which correlates with the excellent 10kHz squarewave response (fig.2), but also that an ultrasonic peak develops quite strongly into 8 ohms (blue and red traces), less so into 4 ohms (cyan, magenta), and not at all into 2 ohms (green). (This resonance results in some ringing that can be seen overlaying the squarewave.) The HF response also rolls off a little earlier into 2 ohms than it does into 4 and 8 ohms. This peak is due to the internal low-pass filter used on the amplifier's output to reduce the level of radio-frequency switching noise that results from the class-D output stage. Even so, I measured 55mV of noise with a center frequency of 595kHz present at the speaker terminals with no input signal present.
Fig.1 NuForce Icon, frequency response at 2.83V into: 8 ohms (left channel blue, right red), 4 ohms (left cyan, right magenta), 2 ohms (green), and simulated loudspeaker load (black). (1dB/vertical div.)
Fig.2 NuForce Icon, small-signal 10kHz squarewave into 8 ohms.
This noise compromises the wideband, unweighted signal/noise ratio to just 36.7dB (ref. 2.83V into 8 ohms). This improved to 70.7dB when the measurement bandwidth was restricted to the audio range, and to 76.5dB when an A-weighting filter was switched in circuit. The Icon is a little noisier than usual for a power amplifier, and I note that the NuForce website recommends against using a loudspeaker with a sensitivity greater than 92dB because the background noise will become audible within 23' of the tweeter. Because of the contaminating effect of this ultrasonic noise, I measured the Icon's distortion and channel separation using the Audio Precision passive low-pass filter. The channel separation (fig.3) was good, at 74dB (RL) and 83dB (LR) at 1kHz and below, these figures decreasing by 15dB at 20kHz.
Fig.3 NuForce Icon, channel separation (LR, blue; RL, red).
Fig.4 shows how the THD+noise percentage in the Icon's output varies with output power. The traces slope down below 200mW or so, indicating that distortion lies beneath the background noise at these low powers. But the distortion remains between 0.1% and 0.2% at higher powers, which is higher than is suggested by the 0.06% specification. Into 8 ohms, the amplifier reaches 1% THD, our usual definition of clipping, at 8.5W (9.3dBW), and doesn't reach its specified maximum output power of 12Wpc (10.8dBW) until just below 3% THD. The 4 ohm trace in this graph stops just above 10W, when the amplifier's protection circuit kicked in. Reducing the signal level allowed the amplifier to turn itself back on again. Clearly, the Icon is not intended to be used with low-impedance speakers, which is why I didn't test its maximum power into 2 ohms. (It turned off at continuous levels above a couple of watts into 2 ohms.)
Fig.4 NuForce Icon, distortion (%) vs 1kHz continuous output power into (from bottom to top at 100mW): 8 ohms, 4 ohms.
Fig.5 plots the THD+N percentage against frequency at 2.83V, an output level equivalent to 1W into 8 ohms (blue and red traces) and 2W into 4 ohms (green, magenta). The lowest distortion can be seen to occur in the upper midrange, with increases both above and below that region. The rise in THD at low frequencies is particularly pronounced into 4 ohms, reaching 1% at 20Hz, which spectral analysis revealed to be due to a large amount of second-harmonic distortion (fig.6). The THD waveform at higher frequencies (fig.7) also has second-harmonic content, but with higher-frequency spikes apparent. Looking at the waveform on a 20MHz analog 'scope indicated that every positive waveform peak was accompanied by a small burst of switching noise, while spectral analysis (fig.8) revealed a regular series of higher-order harmonics as well. The Icon's reduced linearity at high frequencies results in quite high levels of intermodulation distortion when the amplifier is used to drive an equal mix of 19 and 20kHz tones at about half-power into 4 ohms (fig.9). The 8 ohm performance (not shown) was actually a little worse.
Fig.5 NuForce Icon, THD+N (%) vs frequency at 2.83V into: 8 ohms (left channel blue, right red), 4 ohms (left green, right magenta).
Fig.6 NuForce Icon, spectrum of 50Hz sinewave, DC1kHz, at 6.2W into 4 ohms (linear frequency scale).
Fig.7 NuForce Icon, 1kHz waveform at 2W into 8 ohms (top), 0.26% THD+N; distortion and noise waveform with fundamental notched out (bottom, not to scale).
Fig.8 NuForce Icon, spectrum of 1kHz sinewave, DC10kHz, at 6.2W into 4 ohms (linear frequency scale).
Fig.9 NuForce Icon, HF intermodulation spectrum, DC24kHz, 19+20kHz at 3W peak into 4 ohms (linear frequency scale).
I assessed the performance of the NuForce Icon's D/A section by driving the amplifier's USB input with my Macintosh Powerbook and examining the analog output signal at the line-out jack. The Burr-Brown DAC used in the Icon is a 16-bit delta-sigma chip with an integral digital low-pass filter and headphone output, capable of operating up to a sample rate of 48kHz (footnote 1). It offers low linearity error down to 90dB or so (fig.10) and a fairly clean noise floor, though some second-harmonic distortion can be seen with a dithered 1kHz tone at 90dBFS (fig.11). However, the waveform of an undithered sinewave at exactly 90.31dBFS, which should comprise three well-defined DC voltage levels, was obscured by HF noise (fig.12).
Fig.10 NuForce Icon, USB data input, linearity error.
Fig.11 NuForce Icon, USB data input, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 16-bit CD data (right channel dashed).
Fig.12 NuForce Icon, USB data input, waveform of undithered 1kHz sinewave at 90.31dBFS, 16-bit data.
The problem with using USB to get audio data out of a computer to feed to an external DAC is that it is not optimized for uninterrupted streaming. The host PC has operating-system housekeeping chores to attend to, and while the sample rate of the output data, averaged over a longish period, will indeed be 44.1kHz or 48kHz, there will be short-term fluctuations. The Icon's Burr-Brown PCM2706 recovers the audio clock from the USB data packets and uses on-chip analog phase-locked loops to reduce the effects of word-clock jitter. Even so, using the Miller Jitter Analyzer to examine the Icon's analog line-stage output while I fed it the diagnostic signal from my PowerBook gave a high 1.83 nanoseconds of word-clock jitter.
Looking at the spectrum of the analog output signal (fig.13), the noise floor is about 8dB higher than I have measured with the best 16-bit playback systems, and features a large number of discrete noise spikes (blue numeric markers). Jitter sidebands are either data-related (red markers), which are fairly low in level, or of unknown origin (purple markers). Of these, the highest in level lie at ±10.3Hz (purple "1"), ±135Hz (purple "8"), and ±714Hz (purple "19"). There is also some spectral spreading of the central peak evident, due to random low-frequency timing fluctuations. It is difficult to predict the effect this behavior will have on the DAC's sound quality, but I would have expected Wes Phillips to have found bigger audible differences between the Icon and the Musical Fidelity X-DACV3, which has excellent rejection of jitter via its S/PDIF input (see my measurements).
Fig.13 NuForce Icon, USB data input, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 16-bit data. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.
It is easy to be overcritical of an amplifier that costs as little as the NuForce Icon, especially as it includes a digital data input and a headphone output. For what you pay, the engineering compromises managed by its design team seem well arranged for a desktop amplifier. However, I would not recommend it for use in an audiophile's primary system.John Atkinson
Footnote 1: This chip is very similar to the PCM2705 used in the Apple Airport Express and Wadia 170 iTransport, but lacks that chip's USB-to-S/PDIF converter.John Atkinson