Columns Retired Columns & Blogs |
Chord Electronics Mojo D/A headphone amplifier Measurements
Sidebar 3: Measurements
Footnote 1: This test was suggested to me by Jürgen Reis, chief engineer of MBL.
I measured the Chord Mojo with my Audio Precision SYS2722 system (see www.ap.com and the January 2008 "As We See It"). The Mojo's battery was fully charged at the start of the measurements. The data sources were USB from my 2012 MacBook Pro running on battery power, using Pure Music 2.0 to play the files, and S/PDIF on TosLink from the SYS2722. Macintosh's USB Prober utility reported the Mojo's product string as "Mojo" from "Chord Electronics Ltd," and confirmed that its USB port operated in the optimal isochronous asynchronous mode. Apple's AudioMIDI utility confirmed that the Mojo operated at all sample rates from 44.1 through 768kHz with 32-bit integer data. The TosLink input operated reliably up to a sample rate of 96kHz; above that rate, whether the Mojo locked to a datastream sampled at 176.4 or 192kHz depended on the quality of the optical interlink used.
When I first turned on the Mojo in headphone mode, its output level was set to 774.5mV. With both volume buttons illuminated dark blue, the output level was the CD standard's 2V. When I turned on the Mojo in Line mode, the output was fixed at 3.08V. With digital data at 0dBFS, the Mojo's output was 7.425V with the volume control set to its maximum, which is much higher than I expected. However, the waveform was well into clipping. The maximum output level before waveform clipping was 4.887V; except where noted in the text, all measurements were performed with the volume control set so that this output voltage was the reference level.
The Mojo preserved absolute polarity (ie, was non-inverting). The output impedance at low and middle frequencies was extremely low, at 0.7 ohm including the interconnect cable, and though it had increased at the top of the audioband, this was to just 1.8 ohms. The Mojo will have no problem driving even the lowest-impedance headphones.
Tested with a "digital black" 44.1kHz WAV file that included one sample raised to 0dBFS, the Mojo's impulse response (fig.1) revealed the reconstruction filter to be a time-symmetrical FIR typealthough, as with Chord's Rob Wattsdesigned Hugo TT, with more coefficients than I usually find. The ultrasonic rolloff associated with this filter is indicated by the red and magenta traces in fig.2, taken with 44.1kHz-sampled white noise at 4dBFS (footnote 1). The output drops sharply just before the Nyquist frequency (half the sample rate: vertical green line), and other than low-level spurious tones at 31, 62, and 93kHz, the filter's stop-band behavior is very clean. The rolloff is very fast, so with a full-scale tone at 19.1kHz (fig.2, blue and cyan traces), the aliasing product at 25kHz (44,10019,100) is suppressed by >120dB. The harmonics of this tone can also be seen to be very low in level.
Fig.1 Chord Mojo, impulse response at 44.1kHz (4ms time window).
Fig.2 Chord Mojo, wideband spectrum of white noise at 4dBFS (left channel red, right magenta) and 19.1kHz tone at 0dBFS (left blue, right cyan), with data sampled at 44.1kHz (20dB/vertical div.).
Fig.3 shows a more conventional frequency-response graph, taken with data sampled at 44.1, 96, and 192kHz. The overall response conforms to the same basic shape at all three sample rates, with a sharp rolloff just below half the rate at 44.1 and 96kHz. I haven't shown the frequency response with data sampled at 384kHz, as it overlaid the 192kHz response up to 96kHz. Above that frequency, the response followed the same gentle rolloff, reaching 20dB at 165kHz. Channel separation below 1kHz was superb, at >114dB in both directions, and was still >100dB at 20kHz (fig.4). The Mojo's low-frequency noise floor was also superbly clean and free from spurious tones (fig.5).
Fig.3 Chord Mojo, frequency response at 12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel green, right gray), 96kHz (left cyan, right magenta), 192kHz (left blue, right red) (2dB/vertical div.).
Fig.4 Chord Mojo, Channel separation.
Fig.5 Chord Mojo, spectrum with noise and spuriae of dithered 1kHz tone at 0dBFS with 24-bit TosLink data (left blue, right red) (20dB/vertical div.).
Fed dithered TosLink data representing a 1kHz tone at 90dBFS with first 16-bit data (fig.6, cyan and magenta traces), then 24-bit data (blue, red), the increase in bit depth dropped the noise floor by more than 20dB, suggesting resolution of better than 19 bits. However, repeating this test with 24-bit USB data produced a large number of spurious tones and a random noise floor 34dB higher in level (fig.7). While these tones are well below the threshold of hearing, it is still behavior that I don't like to see. By contrast, 24-bit data sourced from an iPad showed a cleaner spectrum and a lower noise floor (fig.8). And with its high resolution and overall low level of noise, the Chord DAC's reproduction of undithered 16-bit data describing a sinewave at exactly 90.31dBFS was exemplary (fig.9), with zero DC offset and the three DC voltage levels well differentiated. With undithered 24-bit data, the Mojo output a well-formed sinewave despite the very low signal level (fig.10).
Fig.6 Chord Mojo, spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with: 16-bit TosLink data (left channel cyan, right magenta), 24-bit TosLink data (left blue, right red) (20dB/vertical div.).
Fig.7 Chord Mojo, spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 24-bit USB data (left channel blue, right red) (20dB/vertical div.).
Fig.8 Chord Mojo, spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 24-bit data sourced from an iPad 2 (left channel blue, right red) (20dB/vertical div.).
Fig.9 Chord Mojo, waveform of undithered 1kHz sinewave at 90.31dBFS, 16-bit data (left channel blue, right red).
Fig.10 Chord Mojo, waveform of undithered 1kHz sinewave at 90.31dBFS, 24-bit data (left channel blue, right red).
Harmonic distortion at 2V into 300 ohms was very low (fig.11), with the second harmonic the highest in level and lying at 110dB (0.0003%). Even with the volume control set to its highest level before clipping into this load, the second harmonic had risen by only 7dB. Intermodulation was also extremely low (fig.12), with the 1kHz difference tone produced by an equal mix of 19 and 20kHz tones lying at 120dB (0.0001%). However, this graph shows some modulation of the Mojo's noise floor. This is very similar to what I found last month with the Apogee Groove, which, like the Mojo, has a much higher maximum output voltage than what would be expected from something powered by or compatible with 5V USB power.
Fig.11 Chord Mojo, spectrum of 50Hz sinewave, DC1kHz, at 0dBFS into 300 ohms with volume control set to 2V output (left channel blue, right red; linear frequency scale).
Fig.12 Chord Mojo, HF intermodulation spectrum, DC30kHz, 19+20kHz at 0dBFS into 300 ohms, 44.1kHz data (left channel blue, right red; linear frequency scale).
Perhaps this modulation is a function of a DC-to-DC converter in the power-supply circuitry, but the effect can also be seen in the narrowband spectral analysis of the Mojo's output while it decoded 16-bit J-Test data (fig.13). The high-order odd harmonics of the low-frequency LSB-level squarewave are all very close to their correct levels (green line). This graph was taken with TosLink data; when I repeated the analysis with USB data (not shown), the noise-floor modulation could still be seenbut, as anticipated from fig.6, the noise floor was a few dB higher than in fig.10. I very much doubt that this behavior will have audible consequences.
Fig.13 Chord Mojo, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit data via TosLink from AP SYS2722 (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.
Overall, the Chord Mojo offers measured performance that is superb for a portable device, and would not be out of place in a high-priced conventional D/A processor.John Atkinson
Footnote 1: This test was suggested to me by Jürgen Reis, chief engineer of MBL.