Headphone Reviews

I measured the CEntrance DACport with an Audio Precision SYS2722 (see www.ap.com and "As We See It" in the January 2008 issue), as well as, for some tests, my Audio Precision System One Dual Domain and the Miller Audio Research Jitter Analyzer. I drove the DACport with the USB 2.0 output of my Intel MacBook running OS10.6 for the testing, playing back WAV files using Bias Peak 6.2. To avoid any problems of noise contamination, I ran the MacBook on battery power for the testing.

The DACport correctly operated with sample rates ranging from 44.1 to 96kHz, including the important 88.2kHz rate. Its maximum output (at 1kHz into 100k ohms) was to specification, at 3.06V. The output was non-inverting (ie, it preserved absolute polarity), and was sourced from a moderately low impedance of 10 ohms. Fig.1 shows the DACport's frequency response at sample rates of 96kHz (blue and red traces) and 44.1kHz (cyan and magenta traces). The response is perfectly flat within the audioband, with a sharp rolloff just before half the sample rate. This graph was taken with the volume control set to its maximum level; channel matching is superb, within 0.02dB. At lower settings of the volume control, the right channel gradually became a little louder than the left, reaching a maximum difference of +0.4dB at 12:00. At even lower levels, the left started to become louder than the right, reaching +0.11dB at 10:00. Given the DACport's affordable price, this departure from perfect volume-control tracking is not unexpected but is okay.

Fig.1 CEntrance DACport, frequency response at –12dBFS into 100k ohms with 96kHz data (left channel blue, right red) and with 44.1kHz data (left cyan, right magenta; 1dB/vertical div.).

Channel separation (fig.2) was also superb, with maximum values of 120dB at 750Hz (R–L) and 250Hz (L–R). It decreased slightly, to 110dB, at low frequencies, which presumably is the effect of the rising power-supply impedance in this region. It also decreased with increasing frequency in the treble, to 84dB L–R and 88dB R–L at 20kHz, due to the usual capacitive coupling between channels. But this is still excellent performance, given that the headphone output jack shares a ground connection between channels.

Fig.2 CEntrance DACport, channel separation (10dB/vertical div.).

For consistency with the digital tests I have performed since 1989, I determine a product's resolution by sweeping the center frequency of a 1/3-octave bandpass filter from 20kHz to 20Hz while the DAC reconstructs a dithered 1kHz tone at –90dBFS with 16- and 24-bit data. The top pair of traces in fig.3 shows the spectral analysis for 16-bit data: the traces peak at exactly –90dBFS and show just the dither noise used to encode the signal. With 24-bit data (middle traces), the noise floor drops by about 13dB, implying that the DACport's resolution is between 18 and 19 bits, at least in the treble. (The amplifier's noise floor is higher in the lower midrange and below than it is in the treble.) The CEntrance has sufficient resolution to decode a dithered 24-bit tone at –120dBFS (fig.3, bottom traces).

Fig.3 CEntrance DACport, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 16-bit data (top) and 24-bit data (middle at 2kHz), and of dithered 1kHz tone at –120dBFS with 24-bit data (bottom at 1kHz). (Right channel dashed.)

Fig.4 repeats the test, this time using a narrowband FFT technique and plotting the spectrum with a linear frequency scale. Again, the increase in bit depth drops the noise floor by 13dB compared with CD data, and again the noise floor rises a little at low frequencies. The 24-bit data also unmask some spurious tones between 2 and 3kHz, but these are still extremely low in level—the highest lies at –130dB, 0.00003%!—so they will have no subjective consequences. These tones persist even with the volume control set to its minimum, so it's possible that they arise from the DACport's DC/DC converters rather than from its digital processing circuitry. What is notable about fig.4, however, is that the noise floor is even lower overall than suggested by CEntrance's measurements on its website—a rare case of the manufacturer's specification actually being pessimistic!

Fig.4 CEntrance DACport, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 16-bit data (left channel cyan, right magenta) and 24-bit data (left blue, right red).

All the DACport's plot of linearity error against absolute level with 16-bit data (fig.5) shows is the effect of the dither used to encode the test signal. Any error is negligible down to –110dBFS; as a result, the DACport's reproduction of an undithered 16-bit tone at exactly 90.31dBFS is excellent (fig.6). Though it is a little noisier than with the YBA D/A processor also reviewed this month (see fig.4 in that review), the three voltage levels described by the data can still be readily resolved. Increasing the bit depth to 24 gives a good, if slightly noisy sinewave (fig.7), and the excellent waveform symmetry demonstrates that while it may be direct-coupled, the CEntrance has negligible DC offset present on its outputs.

Fig.5 CEntrance DACport, linearity error, dBr vs dBFS, 16-bit data.

Fig.6 CEntrance DACport, waveform of undithered 1kHz sinewave at –90.31dBFS, CD data (left channel blue, right red).

Fig.7 CEntrance DACport, waveform of undithered 1kHz sinewave at –90.31dBFS, 24-bit data (left channel blue, right red).

As the DACport's primary purpose is to drive headphones, I performed most of the distortion tests into 300 ohms, which is typical of various Sennheiser models. Fig.8 shows the spectrum of the DACport's output while it drove a full-scale 1kHz tone into that load with the volume control at its maximum: the third harmonic is the highest in level, at just –96dB (0.0015%), with the second harmonic at –100dB (0.001%) in the right channel (red trace), and a little lower in the left (blue). Higher-order harmonics all lie below –116dB (0.00015%), and while the spurious tones still make an appearance just above the noise floor, they lie an octave higher than with the low-level signal shown in fig.5.

Fig.8 CEntrance DACport, spectrum of 1kHz sinewave at 0dBFS into 300 ohms, 24-bit data (left channel blue, right red; linear frequency scale).

Dropping the load impedance to 30 ohms, typical of the Grado models, gave the spectrum shown in fig.9: the DACport is obviously working much harder than into 300 ohms, and not only have the second and third harmonics risen by 20dB or so, many high-order harmonics can be seen. However, this level will be deafeningly loud into low-impedance headphones, and reducing the signal level by 6dB into the same load virtually eliminated the high-order harmonics and dropped the second and third to an innocuous –94dB (0.002%) and –89dB (0.004%), respectively (fig.10). Intermodulation distortion was also very low (fig.11), though in this graph the second-order difference component at 1kHz, resulting from an equal mix of 19 and 20kHz tones, each at –6dBFS, was a little higher in the right channel (red trace) than the left (blue). But unlike the Marantz player reviewed in the July issue, the DACport's digital filter effectively suppresses ultrasonic images of the test tones.

Fig.9 CEntrance DACport, spectrum of 1kHz sinewave at 0dBFS into 30 ohms, 24-bit data (left channel blue, right red; linear frequency scale).

Fig.10 CEntrance DACport, spectrum of 1kHz sinewave at –6dBFS into 30 ohms, 24-bit data (left channel blue; linear frequency scale).

Fig.11 CEntrance DACport HF intermodulation spectrum, 19+20kHz at 0dBFS peak into 300 ohms, 24-bit data (left channel blue; linear frequency scale).

Finally, while it operates in the potentially jitter-prone isochronous USB mode, the DACport's analog output is superbly free from jitter-related artifacts. Fig.12, for example, was taken with 16-bit J-Test data; not only are the harmonics of the low-frequency squarewave at the residual level, the central spike representing the Fs/4 sinewave shows almost no spectral spreading, implying that the DACport's circuit effectively rejects low-frequency random jitter. The Miller Analyzer estimated the jitter to be just 91 picoseconds peak–peak, which is below its effective resolution limit. With 24-bit J-Test data, the DACport's output spectrum was superbly clean (fig.13), without a trace of sidebands at 1kHz, the USB sample-rate polling frequency.

Fig.12 CEntrance DACport 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, 11kHz; frequency range, ±3.5kHz (left channel blue, right red).

Fig.13 CEntrance DACport high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 24-bit data. Center frequency of trace, 11kHz; frequency range, ±3.5kHz (left channel blue, right red).

Yes, it is possible to get DACs with slightly lower noise floors, but the CEntrance DACport demonstrates superb audio engineering in both the digital and analog domains, which is even more commendable considering that it has to get its power from the host computer and operates in USB's adaptive mode.—John Atkinson